Fig 1.
Integration of experimental data from MR in dogs and VO in rats.
Each plot shows fold changes in one reported measure, plotted as a function of time normalized by the time constant of an exponential equation fitted to the LVM/BW data. a) LVM/BM data from sixteen articles (41 datasets) on experimental MR in dogs [15,16,20–33] and seventeen articles (45 datasets) on experimental VO in rats [34–50]. b) 6 measurements of serum angiotensin II (ANGII) from three articles on experimental MR in dogs [29,30,51] and 7 measurements from six articles on experimental VO in rats [38,41–43,52,53]. c) 3 measurements of serum atrial natriuretic peptide (ANP) levels from three studies in dogs [51,54–56] and 9 measurements from five studies in rats [39,53,57–59]. d) Norepinephrine (NE): 10 datasets from four MR studies in dogs [21,31,60,61] and 7 datasets from six studies in rats [39,44,48,52,59,62]. e) Endothelin 1 (ET1): two datasets from two studies in dogs [63,64] and two datasets from two studies in rats [36,42]. f) Brain natriuretic peptide (BNP): three datasets from three articles on experimental canine MR [51,54–56]; two datasets from two articles on experimental VO in rats [38,57]. Probability distribution functions (PDFs) are derived from fits to equations listed in Table C in S1 Text. PDFs are presented in shaded gray, median in solid line and quantiles in dashed lines.
Fig 2.
Effects of inflation and growth on end-diastolic stretch relative to unloaded state in a spherical model and in hearts fixed at control volumes.
a) Spherical model predicts that stretch increases with inflation (black solid line), while eccentric hypertrophy shifts the stretch-volume curve downwards, to lower stretch at any given volume (red dashed line). b) Experimental measurements of sarcomere stretch in dog hearts fixed at controlled volumes similarly show increased stretch between physiological loading conditions (Baseline) and acute volume overload (Acute dilation), and a shift in the stretch-volume relationship following 10% growth in LVM during long-term volume overload. Figure generated based on data from Ross et al. [66].
Fig 3.
Changes in volume, mass, and end-diastolic strain during experimental MR in dogs.
a) Fitted data on fold changes show a) a 36% increase in LVM [15,16,20–33] and b) 67% increase in LV end-diastolic volume (VED) [16,17,20–26,29,32,33,61,78,80,82,83,131–136] on average at steady-state. c) In a spherical model, these changes imply a 20% increase followed by a gradual decrease below baseline in end-diastolic strain on average, with a wide range of trajectories possible for different combinations of mass and volume change. d) Calculated chronic strain levels are lower when LVM increases more. Insets in panels (a-c) show acute changes plotted on a magnified time axis.
Fig 4.
Predicted chronic activation of the cardiomyocyte signaling network during VO.
Colors indicate the logarithm of the fold change in activity of each node in chronic stages of VO relative to baseline, with red indicating an increase and blue indicating a decrease.
Fig 5.
Results of MC simulations of VO and agonist infusion.
Top panels show normalized growth in CellArea for 100 representative simulations (gray lines), with terminal values indicated by black circles; adjacent boxplots reflect the mean, CI50, and CI90 of terminal values for 1000 simulations of a) Untreated VO, c) ISO infusion, and e) AngII infusion. Bottom panels contain heatmaps comparing the predictions of changes in signaling network outputs (SERCA, αMHC, βMHC, ANP, BNP, and CellArea) to experimental data following b) Untreated VO, d) ISO infusion, and f) AngII Infusion. Red indicates that more than 75% of simulations predicted an increase in the output, or that the majority of studies reported a significant increase. Blue indicates a decrease in >75% of simulations or the majority of experiments. Gray indicates the absence of consistent changes in simulations or significant changes in experiments. White numbers indicate the number of studies reporting a significant change and the total number of studies reporting data for each output.
Fig 6.
Validation of model predictions of growth during infusion of hypertrophic agonists with and without receptor blockers.
a) Boxplot shows the mean, 50% confidence interval, and 90% confidence interval for predicted normalized growth in 1000 MC simulations of ISO infusion with the calibrated network. Twelve of 15 experimental means (plotted as individual points with error bars) from [86–97] fell within the CI90. b) Model-predicted CI90 for growth during AngII infusion encompassed most published values at the earliest time point and all reported experimental means at later time points [98–111]. c) Simulations of four NE infusion experiments in dogs with different doses, fold increases in plasma concentrations (Conc.), and termination times showed reasonable agreement with the limited available data [112–115]. d) Simulations of agonist infusion plus receptor blockers replicated reported significant effects of β blockers (βB) on ISO-induced hypertrophy infusion [88,95,116,117], and smaller effects that were significant in some but not all studies for ARB administration during ISO infusion [86,96,97,117,118] or ERA administration during AngII infusion [111,119,120]. By contrast, the model prediction of nearly equal likelihood that ERA increases or decreases NE-induced growth appears to contradict a significant reported decrease in the only available experimental study [121]. Red coloring of boxplots indicates increases in >75% of simulations, blue indicates decreases in more than 75%, and gray indicates mixed results. Points with error bars indicate experimental mean±SD for studies reporting statistically significant effects (filled markers) or non-significant effects (open markers), from studies in dogs (circles) or rats (triangles).
Fig 7.
Effects of receptor blockers on VO-induced hypertrophy in published experiments and the calibrated model.
a) Simulations of β blockers (βB) at both early and late stages
of VO, simulations (boxplots) were equally likely to predict increases or decreases, with no net effect on average. MR experiments in dogs (circles) [15,16,80] and VO in rats (triangles) [14,124] also found no significant effect early but a further increase in growth at later timepoints relative to untreated VO. b) Simulated administration of angiotensin receptor blockers (ARB) also produced mixed results, consistent with mixed results in published studies [29,42,46,50]. By contrast, simulated ERA administration decreased VO-induced hypertrophy in the majority of simulations, and in all four available experiments [125–128]. Blue coloring of boxplot indicates predicted decrease in >75% of simulations, gray boxplots indicate mixed simulation results, filled markers indicate experimental means associated with statistically significant effects and open markers indicate means associated with non-significant effects.
Fig 8.
Sensitivity analysis of network outputs to calibrated network inputs in chronic stages of VO.
Each marker represents the final state of one of 1,000 Monte Carlo simulations of VO. A linear regression model (dashed line) is fitted to each output-input pair, and the Pearson PCC is displayed on the upper left corner of each. Among the inputs known to vary substantially over the course of VO, ET1 displayed the largest influence on all outputs in the late stages of overload.