Apparent growth tensor of left ventricular post myocardial infarction – In human first natural history study

https://doi.org/10.1016/j.compbiomed.2020.104168Get rights and content

Highlights

  • Growth tensors are estimated from left ventricle cine MRI and late gadolinium enhanced images post myocardial infarction.

  • Patients can be classified as Dilation, No-Change and Shrinkage groups; with wall remodelling in all groups.

  • The growth tensor invariants can be used as biomarkers for the growth and remodelling processes.

Abstract

An outstanding challenge in modelling biomechanics after myocardial infarction (MI) is to estimate the so-called growth tensor. Since it is impossible to track pure growth induced geometry change from in vivo magnetic resonance images alone, in this work, we propose a way of estimating a surrogate or apparent growth tensor of the human left ventricle using cine magnetic resonance (CMR) and late gadolinium enhanced (LGE) images of 16 patients following acute MI. The apparent growth tensor is evaluated at four time-points following myocardial reperfusion: 4–12 h (baseline), 3 days, 10 days and 7 months. We have identified three different growth patterns classified as the Dilation, No-Change and Shrinkage groups defined by the left ventricle end-diastole cavity volume change from baseline. We study the- trends in both the infarct and remote regions. Importantly, although the No-Change group has little change in the ventricular cavity volume, significant remodelling changes are seen within the myocardial wall, both in the infarct and remote regions. Through statistical analysis, we show that the growth tensor invariants can be used as effective biomarkers for adverse and favourable remodelling of the heart from 10 days onwards post-MI with statistically significant changes over time, in contrast to most of the routine clinical indices. We believe this is the first time that the apparent growth tensor has been estimated from in vivo CMR images post-MI. Our study not only provides much-needed information for understanding growth and remodelling in the human heart following acute MI, but also identifies novel biomarker for assessing heart disease progression.

Introduction

Myocardial infarction (MI), also known as heart attack, involves cardiomyocyte death following the loss of antegrade coronary blood flow. After acute MI, the left ventricle (LV) undergoes a sequence of adaptations in structure and function, regulated by mechanical, neuro-hormonal and genetic factors [7,17,47]. This myocardium growth & remodelling (G&R) may become maladaptive. The LV may experience dilation in association with wall thinning around the infarct zone, and hypertrophy or fibrosis (collagen fibre scar) in the remote zone. Adverse G&R, i.e. dila-tion, can lead to heart failure, a major source of morbidity and mortality [16], and sudden cardiac death due to malignant arrhythmias. Over the years, with improvement in the treatment of acute MI, more people are surviving post-MI, but with a subsequent rise in heart failure [42,51].

Cardiac G&R has previously been categorized into two types: (1) concentric G&R with wall thickening caused by pressure over-loading; (2) eccentric G&R with dilation of the ventricular cavity and wall stretching, both induced by pressure over-loading [3,24]. However, MI-induced LV G&R is more complex, and recently a classification, such as reverse remodelling, no remodelling, and adverse remodelling, was proposed in terms of 12% relative changes in LV end-diastolic volume (LVEDV) [11]. MI-induced progressive G&R is an important clinical determinant of heart failure [30], and arresting or prompting reverse remodelling is a long-term aim to reduce the risk of heart failure [7].

Usually, LV G&R is quantified by evaluating LV global changes, such as LVEDV, end-systolic volume (ESV), ejection fraction (EF) and wall thickness [10]. However, these indices only describe an incomplete picture of the G&R process. More detailed local patterns of G&R in MI patients, the difference in MI and remote regions can provide additional insights into identifying patients with a high risk of adverse remodelling and evaluating interventions aimed at reducing the progression of heart failure.

Quantification of local LV G&R following MI is difficult, requiring serial measurements of in vivo myocardium geometry and motion. For research purposes, measurements are often made in animals by tracking several small-sized beads embedded in the LV wall through open-chest surgery [26,27,39,49]. For example, Kass et al. [27] measured disproportionate epicardial dilation after transmural infarction of the canine LV within 7 days of acute MI. By suturing stainless steel spherical beads to the epicardial surface, they found that 1-h after infarction, the end-diastole area in the infarct region increased by 20.3% above healthy control (normal LV before infarction) compared with a 7.9% increase in the remote region. 24-hour post-MI, both the regions expanded by an additional 10%. One-week post-MI, the end-diastole area in the infarct region increased to 31.4% above the control, but the remote area became to only 8.5% above the control. Holmes et al. [26] measured scar transmural remodelling after infarction in five porcine hearts. Fitting the measured bead positions into a finite element (FE) model at 1- and 3-week post-MI, they calculated the post-MI strain against the reference configuration (at end-diastole before infarction). Their results showed that at 1-week post-MI, infarct expansion dominated the remodelling process, as characterized by the circumferential and longitudinal in-plane stretch and wall thinning. However, post-MI strain at 3-week is much more complex. Two out of five LVs showed continuous in-plane infarction expansion with wall-thinning, and the remaining three had in-plane shrinkage either with wall thinning or thickening. In a more recent study, Tsamis et al. [49] implanted three transmural columns of bead sets across the mid-lateral equatorial region of a sheep LV and recorded the bead positions using biplane cine-radiography before infarction and 7-week post-MI. They calculated the end-diastolic and end-systolic post-MI strains in a lateral-equatorial region adjacent to infarct by fitting bead positions using a FE model and found there were chronic muscle fibre lengthening (> 10%) in the longitudinal direction, 25% shrinkage in the radial or transmural directions, and 15% decrease in the ventricular volume.

Although animal experiments can provide useful data for calibrating, validating and verifying various cardiac growth modelling theories, the bead implantation method is invasive and not suitable for human subjects. With the development of cardiac magnetic resonance (CMR) [5,6,15,54] and ultrasound imaging [45,46], it is now possible to quantify some regional geometry and function post-MI in vivo. However, using in vivo CMR images to track G&R post-MI is challenging. Longitudinal 3D cine images may not be available, motion artefacts may affect in the images, and ventricular geometries at different time-points must be co-registered. O'Regan et al. [40] studied LV remodelling during the first year after reperfused acute ST-elevation MI (STEMI) in 47 patients using a three-dimensional (3D) co-registration approach. Both short-axis cine and late gadolinium enhanced (LGE) images were co-registered to an atlas template and local expansion of the heart wall was tracked by using intensity-based similarities. End-diastolic remodelling is defined as the mean change of vertex separations from the matched endocardial and epicardial surfaces. Their results showed that in-plane expansion in the infarct region is greater than in the remote myocardium (1.6%±1.0 vs 0.3%±0.9), and associated with wall-thinning in the infarct region but no changes in wall thickness in the remote region. Additionally, they demonstrated that MI transmurality and microvascular obstruction (MVO) may affect local remodelling considerably. To our knowledge, O'Regan et al. [40] were the first group to report patient measurements of MI-induced local myocardial remodelling in vivo. However, they only studied the local expansion characterization and did not estimate the growth tensor.

Understanding G&R of the LV post-MI is critically important to risk stratifying preventive therapy for heart failure. A key element of monitoring a diseased heart is to extract growth tensor evolution with time from non-invasive clinical images. Computational models based on continuum mechanics and clinical imaging assessment techniques have recently been developed to overcome the limitations by describing G&R based on routine clinical images [2,9,55,56].

In this work, to quantify G&R in patients with recent MI and evaluate its relation to adverse remodelling, a continuum mechanics approach is employed to evaluate the LV apparent growth tensor in 16 patients post-MI undergoing CMR at 4 distinct time-points: at 4–12 h (baseline), 3 days, 10 days, and 7 months after reperfusion. Primary percutaneous coronary intervention (PPCI) was performed within a few hours (<12 h) following the onset of acute MI. The end-diastolic (ED) LV geometries are first reconstructed based on CMR cine images at different time points. The MI regions are then delineated from corresponding short-axial LGE images. The reconstructed LV geometries are co-registered to a template LV finite element mesh, and the apparent growth tensors at end-diastole in MI and remote regions are calculated against the baseline LV geometry. All patients are further classified into three groups (Dilation, Shrinkage or No-Change) according to their end-diastolic volume changes based on the criteria proposed by Bulluck et al. [11] for adverse remodelling, reverse remodelling and no-remodelling behaviours. Finally, the invariants of the growth tensor extracted in both the MI and remote regions are investigated and their statistical significance is examined against the short-term (one week) and long-term (6 months) pump function recovery in these groups. We believe the outcomes of this study will be useful to quantitatively characterize G&R for patients post-MI, and provide much needed clinical input for mechanobiology modelling of G&R process in the human heart.

Section snippets

Patient population and CMR protocol

This study involves 16 STEMI patients with four longitudinal CMR scans, at 4–12 h, 3 days, 10 days, and 7 months after reperfusion. All patients were enrolled between May 2011 to November 2012 with acute STEMI (Clinical trial registration no, NCT 02072850). All patients provided written informed consent.

CMR imaging was performed on a Siemens MAGNETOM Avanto 1.5-T scanner (Erlangen, Germany) using an anterior phased-array body coil (12-element) and a posterior phased-array spine coil

Validations

Validation 1: the LVEDV and the MI volume ratio (VMI/Vw) from our reconstructions are compared with the measurements obtained by K.M. at 4–12 h and 7 months post-MI, as shown in Fig. 7. In general, good agreement of the LVEDV volume ratio is observed between clinical assessment and values derived from our 3D reconstructions with a mean error (clinical - reconstruction) of +14.43% at 4–12 h and +11.13% at 7 months, respectively. Excellent agreement of the MI volume ratio is also achieved with a

Results

Fig. 9 shows the normalized LVEDV changes (LVEDV/LVEDV0) at 3 days, 10 days and 7 months after reperfusion. According to the criterion of a ±12% LVEDV change at 7-months post-MI [11], out of 16 patients, nine (ID: 1, 2, 3, 4, 6, 7, 8, 11, 12) (56.25%) are in the Dilation group; two (ID: 9, 10) (12.5%) are in the No-Change group, and five (ID: 5, 13, 14, 15, 16) are in the Shrinkage group. The averaged clinical indices at baseline and 7-month follow up for the three groups are listed in Table 3.

Discussion

To the best of our knowledge, this has been the first time that growth tensor is quantitatively analysed from the human studies post-MI. It is challenging to measure growth tensors in situ from animal studies [26,49], but more so from in vivo clinical studies. Imaging techniques can therefore provide an alternative source of information. For example, O'Regan et al. [40] quantified LV remodelling by tracking ventricular expansion using a 3D co-registration approach based on intensity-based

Conclusion

In this paper, we have developed an image-derived method for estimating apparent myocardial growth tensors in 16 patients post-MI. Both routinely available cine and LGE CMR images are used for personalized LV geometry reconstruction. We have demonstrated that the estimated growth tensor agrees well with published experimental data and conventional clinical observation. Our results show that the greatest shape change occurs in the Dilation group and the least in the Shrinkage group. Both Gff and

Funding

This project is funded by the UK Engineering and Physical Sciences Research Council grants (EP/S020950, EP/S014284/1, EP/S030875, and EP/N014642).. KM and CB were also supported by grants from the British Heart Foundation (PG/11/228474; PG/14/64/31043; FS/15/54/31639; CoE/RE/18/6134217).

Data availability statement

All quantities including the growth tensor estimated for all 16 patients are made available through the Supplemental Data.

CRediT authorship contribution statement

Wenguang Li: Segmentation, Reconstruction, Software, improvement, Data curation, Result analysis, Writing - original draft. Hao Gao: Methodology, Software, Data curation, Critical reviewing. Kenneth Mangion: Data curation, Clinical analysis, Critical reviewing. Colin Berry: Data curation, Clinical analysis, Critical reviewing. Xiaoyu Luo: Conceptualization, of this study, Result analysis, Critical reviewing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

We would like to thank Prof. Ray Ogden for helpful discussions.

References (56)

  • V.K. Sudarshan et al.

    Data mining framework for identification of myocardial infarction stages in ultrasound: a hybrid feature extraction paradigm (part 2)

    Comput. Biol. Med.

    (2016)
  • A. Tsamis et al.

    Kinematics of cardiac growth: in vivo characterization of growth tensors and strains

    J. Mech. Behav. Biomed. Mater.

    (2012)
  • H. Yousefi-Banaem et al.

    Prediction of myocardial infarction by assessing regional cardiac wall in cmr images through active mesh modelings

    Comput. Biol. Med.

    (2017)
  • D. Ambrosi et al.

    Growth and remodelling of living tissues: perspectives, challenges and opportunities

    J. R. Soc. Interface

    (2019)
  • S. Ardekani et al.

    Computational method for identifying and quantifying shape features of human left ventricular remodeling

    Ann. Biomed. Eng.

    (2009)
  • J. Arumugam et al.

    Model of anisotropic reverse cardiac growth in mechanical dyssynchronyn

    Sci. Rep.

    (2019)
  • S.F. Baracho et al.

    A segmentation method for myocardial ischemia/infarction applicable in heart photos

    Comput. Biol. Med.

    (2017)
  • A.S. Bhatt et al.

    Adverse remodeling and reverse remodeling after myocardial infarction

    Curr. Cardiol. Rep.

    (2017)
  • A. Bône et al.

    Learning distributions of shape trajectories from longitudinal datasets: a hierarchical model on a manifold of diffeomorphisms

  • H. Bulluck et al.

    Cardiovascular magnetic resonance in acute st-segment–elevation myocardial infarction: recent advances, controversies, and future directions

    Circulation

    (2018)
  • H. Bulluck et al.

    Defining left ventricular remodeling following acute ST-segment elevation myocardial infarction using cardiovascular magnetic resonance

    J. Cardiovasc. Magn. Reson.

    (2017)
  • J. Chen et al.

    Remodeling of cardiac fiber structure after infarction in rats quantified with diffusion tensor mri

    Am. J. Physiol. Heart Circ. Physiol.

    (2003)
  • J. Duan et al.

    Automatic 3d bi-ventricular segmentation of cardiac images by a shape-refined multi-task deep learning approach

    IEEE Trans. Med. Imag.

    (2019)
  • J.J. Gajarsa et al.

    Left ventricular remodeling in the post-infarction heart: a review of cellular, molecular mechanisms, and therapeutic modalities

    Heart Fail. Rev.

    (2011)
  • H. Gao et al.

    Changes and classification in myocardial contractile function in the left ventricle following acute myocardial infarction

    J. R. Soc. Interface

    (2017)
  • H. Gao et al.

    Left ventricular strain and its pattern estimated from cine CMR and validation with DENSE

    Phys. Med. Biol.

    (2014)
  • H. Gao et al.

    Dynamic finite-strain modelling of the human left ventricle in health and disease using an immersed boundary-finite element method

    IMA J. Appl. Math.

    (2014)
  • H. Gao et al.

    Parameter estimation in a Holzapfel–Ogden law for healthy myocardium

    J. Eng. Math.

    (2015)
  • Cited by (11)

    • Physics-informed graph neural network emulation of soft-tissue mechanics

      2023, Computer Methods in Applied Mechanics and Engineering
    • Electro-fluid-mechanics of the heart

      2022, Journal of Fluid Mechanics
    View all citing articles on Scopus
    View full text