T-World Virtual Human Cardiomyocyte. II. Organ-Scale Simulations and Applications

Figures

• 

  Figure 1.. Organ-level arrhythmia development and contraction. A, Healthy biventricular model constructed from clinical magnetic resonance imaging data and ECG simulation (solid line) compared with the ECG record from the patient used for the ventricular anatomy. B, Ventricular fibrillation simulation in the setting of acute ischemia when the stimulation rate is progressively increased. Heart snapshots above the ECG illustrate the spatial distribution of membrane potentials at different stages of progression toward fibrillation. Examples of action potentials in normal, border, and infarct zones before transition to arrhythmia are given in Figure S1. C, Snapshots of a contracting heart built using T-World during (1) diastole, (2) early contraction during electrical activation, and (3) end-systolic contraction. D, Corresponding pressure-volume loop, and pressure and volume transients. Videos of transition to ventricular fibrillation and contraction are provided as the Supplemental Material (Videos S1 and S2).

• 

  Figure 2.. In silico drug safety and efficacy assessment. A, Schematic of the in silico drug trial procedure, showing how pharmacological data on dose-dependent inhibition of various currents by cardioactive drugs are applied to calibrated populations of models. Subsequently, arrhythmogenic behaviors (early afterdepolarizations [EADs] and repolarization failure) are detected and scored, and the prediction can be compared with reference clinical risk. B, Predicted risk scores for 60 drugs with T-World, with color-coded clinical risk, as in the study by Passini et al10 and Tomek et al.25 Tables of true/false classifications are provided in the right part for T-World and ToR-ORd. Please see the Methods section for a summary of how several data updates led to a subtly different performance of ToR-ORd in our study compared with the original publication.25 TdP indicates Torsades de Pointes.

• 

  Figure 3.. Detection of problematic drug description data based on phenotypic disagreement. A, Concentration-dependent effect of the first lidocaine description (with INaL effect) on the default T-world action potential (AP; left) and the effect of 100× on the population of models (right), showing overall safety (no model in the population manifests an early afterdepolarization [EAD]). B, Similar to A for the second lidocaine description available in the database (no INaL block), showing gradual dose-dependent AP duration (APD) prolongation to the left and repolarization abnormalities to the right. C, Simulated concentration-dependent effect of cilostazol as described by Kramer et al27 (IC50 for IKr=13.8 µmol/L; IC50 for ICaL=91.2 µmol/L) on AP (left) and contractility (middle), showing AP prolongation, but no marked increase in contractility, and an absence of EADs with 100× in the population of models (right). D, Similar to C but including cilostazol-induced increases in ICaL based on the study by Matsui et al28 (+10% at 10× dose=1.28 µmol/L, +22% at 30× dose, and +40% at 100× dose). This recapitulates the positive inotropic effect of the drug, as well as the proarrhythmic risk. ICaL…

• 

  Figure 4.. Effects of mexiletine (MEX) on action potential duration (APD) and early afterdepolarization (EAD) vulnerability. A, APD shortening by 2 distinct simulated MEX formulations (Supplemental Methods) compared with data in the study by Crotti et al.29 B, Reduction in EAD risk for both MEX formulations (quantified using an EAD threshold: the largest IKr multiplier sufficient to trigger EAD in a long QT syndrome type 2 [LQTS2] version of T-World at 0.25 Hz pacing). C and D, Comparison of EAD thresholds in the 2 MEX formulations, comparing an untreated LQTS2 cell, a cell treated with the simulated MEX formulation, and then a range of MEX treatments where the effects on individual ionic currents were turned off. An increase in those knockout bars compared with the full MEX effect indicates a protective effect of the given channel-blocking effect (omitting it makes EADs appear at a higher IKr level).

• 

  Figure 5.. Arrhythmogenesis promotion by type 2 diabetes (T2D). A, An overview of 6 different versions of a diabetic model, reflecting heterogeneity of literature on T2D remodeling. D1 is the current best-estimate model of ionic remodeling in T2D, D2 represents additional Ca2+/calmodulin-dependent protein kinase II (CaMKII) activity, and D3 to D6 are derived models representing a reduction, rather than an increase in ICaL (see Supplemental Methods for details and supporting literature). B, Comparison of steady-state action potentials of models, D1 to D6. C, Threshold for early afterdepolarizations (EADs) across control and diabetic models (the highest IKr availability, which supports EAD formation at 0.25 Hz pacing). Higher values indicate greater vulnerability to EADs. D, Similar to C when single elements of T2D remodeling considered in D1 to D6 are added to the control model. The +CaMKII column also involves an increase in INaL, as described in the Methods section, and −ICaL corresponds to 80% ICaL density compared with the control model. E, Alternans threshold (the longest pacing cycle length, which induces calcium-transient alternans) between the population of control vs T2D models. The…

• 

  Figure 6.. NaV1.8 current as driver of early afterdepolarizations. A, Activation and inactivation curves of the standard cardiac INa current through NaV1.5 vs right-shifted curves carried by the predominantly neuronal NaV1.8 (see Supplemental Methods for details).40,41 B, Action potentials (APs) of control T-World vs T-World during 1 Hz pacing with 2 different amounts of NaV1.8 current added (expressed as relative percentage of peak INa). C, NaV1.8 current profiles during the APs in B. D, Concurrent visualization of NaV1.8 current and ICaL, demonstrating that the sodium current reactivation precedes ICaL.