Subject-Specific Modeling of the Early Childhood Consequences of Preterm Birth

Introduction: : Neonatal care has improved over the last several decades, leading to an increased number of moderately to extremely preterm ( < 32 weeks gestation) babies surviving into adulthood, with the consequence that we are only now beginning to understand the short- and long-term cardiopulmonary effects of preterm birth [1,2]. The combination of the preterm newborn’s underdeveloped cardiopulmonary system and their frequent need for life-saving respiratory support places them at high risk for cardiopulmonary injury. Through adulthood, those born prematurely are at an increased risk for systemic and pulmonary hypertension, with an up to 17-fold increased risk for heart failure, cardiometabolic diseases, and stroke [3].

Rodent models of preterm birth recapitulate biventricular and multiscale (i.e., including cell- and whole organ-level) changes and are useful tools in studying this disease state [4,5]. However, it can be challenging to synthesize these disparate measurements across multiple organ systems and scales. In-silico, mechanistic modeling, when based on multiscale experimental data, enables robust hypothesis-testing and hypothesis-generation, which can provide insights into the underlying mechanisms of the observed cardiopulmonary dysfunction and simulate physiological outputs that are beyond the available data. Furthermore, subject-specific modeling can provide insight into disease progression in a single individual and generate an enhanced set of biomarkers for contrasting control and disease groups or identifying new disease subgroups.

In this study, we tested our hypothesis that in-silico, subject-specific modeling of biventricular mitochondrial, myofiber, and organ-scale function and circulatory function could capture key features of cardiopulmonary abnormalities of the preterm born in early childhood.

Materials and

Methods: : We calibrated a multiscale model to subject-specific, biventricular pressure-volume timeseries data from a hyperoxic rat model of preterm birth and normoxic controls to investigate the impact of preterm birth on multiscale cardiopulmonary function at a timepoint representing early childhood (postnatal day 21; P21). The animal model was described and the data used was collected in [4].

Our adapted model [6,7] includes four sub-models as summarized in Figure 1: crossbridge kinetics, myofiber mechanics, biventricular mechanics, and circulation; with model inputs: sarcoplasmic reticulum calcium flux, which drives active force generation, and mitochondrial metabolites, which determine transition rate constants between states in the crossbridge cycle. The TrigSeg model simulates interventricular dynamics using three semi-spherical thick-walled segments for the left, right, and septal free walls [8]. In total there are 54 parameters and 46 differential equations, including 33 for crossbridge mechanics, 3 for myofiber mechanics, 4 for TriSeg, and 6 for circulation dynamics.

Heart and circulation parameters were estimated from rat bodyweight and ventricular wall weight ratios, heart rate, end-diastolic and end-systolic pressure and volume data from normoxic control and preterm animal models. Crossbridge parameters were modified to match early childhood (P21) normoxic and hyperoxic Sprague Dawley rats based on studies by [9] and [5].

We conducted a sensitivity analysis to identify which organ-scale parameters were most influential on biventricular pressure and volume timecourses and established a 10-parameter subset for estimation via global optimization. We predicted metrics such as myofiber power and septal curvature, and we determined relationships between clinical features and model parameters.

Results, Conclusions, and Discussions:: The calibrated pressure-volume loop simulations had good agreement with the data, with the model simulations within the bounds of variability seen in the 20 or more heartbeats of data. Of the 10 calibrated parameters, pulmonary arterial resistance, RV and septal wall area, and septal wall volume were significantly increased in hyperoxia. This corresponds with clinical observations of RV dilation and pulmonary hypertension in preterm children.

RV myofiber power for the hyperoxic group was significantly higher than for the normoxic group (p=0.025; hyperoxia: 2.4 ± 0.8, normoxia: 1.6 ± 0.5), and there was no significant difference between the groups for LV myofiber power (p=0.27). This increased RV myofiber power corresponds with an increase in RV relative sarcomere shortening during ejection. The absolute RV myofiber length averaged over the cardiac cycle was lower in hyperoxia than in normoxia; therefore, the increased contractile force was not due to the Frank-Starling mechanism.

Model simulations revealed abnormal septal bounce behavior during systole in hyperoxia characterized by a rapid decrease in curvature, or flattening, of the septum with a subsequent rapid return to normal curvature. This curvature drop was over 10x larger in hyperoxia than in normoxia. Further, from a Pearson’s Correlation, septal bounce was highly significant and moderately linearly related to RV cardiac power (ρ=0.67) and mean pulmonary arterial pressure (ρ=0.72). This suggests that septal bounce may be a non-invasive marker of RV hypercontractility and pulmonary hypertension in preterm birth. Additionally, four hyperoxic subjects with lower pressures compared to other hyperoxic subjects were identified, and when they were removed the correlations were stronger.

Overall, we confirmed our hypothesis that the model can represent RV and pulmonary vascular adaptations in preterm birth consistent with experimental and clinical findings at a timepoint that corresponds to childhood. Our investigations highlight the multiscale nature of cardiovascular consequences of preterm birth, generate testable hypotheses regarding mechanisms and markers of a hypercontractile RV in the preterm born, and provide a framework for further exploration of cardiopulmonary disease manifestation into adulthood. Future work will include creating patient-specific simulations at P90 to understand the progression of cardiopulmonary dysfunction in the preterm birth.

Acknowledgements and/or References (Optional):: Acknowledgements: We acknowledge Daniel A. Beard from the University of Michigan and his contributions.

References:
1. Bavineni, M., et al., Mechanisms linking preterm birth to onset of cardiovascular disease later in adulthood. Eur Heart J, 2019. 40(14): p. 1107-1112.
2. Perez, A., et al., Lost in Transition: Health Care Experiences of Adults Born Very Preterm—A Qualitative Approach. Frontiers in Public Health, 2020. 8.
3. Carr, H., et al., Preterm Birth and Risk of Heart Failure Up to Early Adulthood. Journal of the American College of Cardiology, 2017. 69(21): p. 2634-2642.
4. Kumari, S., et al., Bimodal right ventricular dysfunction after postnatal hyperoxia exposure: implications for the preterm heart. Am J Physiol Heart Circ Physiol, 2019. 317(6): p. H1272-h1281.
5. Patel, J.R., et al., Altered Right Ventricular Mechanical Properties Are Afterload Dependent in a Rodent Model of Bronchopulmonary Dysplasia. Front Physiol, 2017. 8: p. 840.
6. Marzban, B., R. Lopez, and D.A. Beard, Computational Modeling of Coupled Energetics and Mechanics in the Rat Ventricular Myocardium. Physiome, 2020.
7. Kim, S.M., et al., Computational modeling of ventricular-ventricular interactions suggest a role in clinical conditions involving heart failure. Frontiers in Physiology, 2023. 14.
8. Lumens, J., et al., Three-wall segment (TriSeg) model describing mechanics and hemodynamics of ventricular interaction. Ann Biomed Eng, 2009. 37(11): p. 2234-55.
9. Prakash, Y.S., et al., Comparison of cross-bridge cycling kinetics in neonatal vs. adult rat ventricular muscle. Journal of Muscle Research & Cell Motility, 1999. 20(7): p. 717-723.