In Silico Cardiac Safety Modelling

Transform ion channel IC50 values into mechanistic pro-arrhythmia risk assessments to inform clinical development.

Traditional cardiac safety assessment relied on hERG (IKr) screening which can result in false alarms and misses real risks. In silico modelling addresses this by integrating effects across several cardiac ion channels into validated human ventricular myocyte models. 

In silico modelling can transform ion channel IC50 data into mechanistic risk predictions by calculating biomarkers for Torsades de Pointes risk (Figure 1) including qNet (defined as total net charge during one cycle length) and APD90 (action potential duration). This approach is aligned with the 2022 ICH E14/S7B Q&A which recognised the use of in silico models as an optional follow-up tool for TdP risk characterisation and animal reduction under 3R principles.

insilico 1 | ApconiX

Figure 1: In silico simulations, use ion channel IC50 data as input to simulate a compounds effect on the cardiac action potential and predict TdP risk. 

Insilico hero scaled | ApconiX

Benefits of early in silico cardiac screening: 

  • No extra lab work – simulations can use existing IC>₅₀ data as input.
  • Reduces compound attrition – avoids false positives from hERG-centric screening
  • Flags risks early in development - before expensive in vivo or clinical work
  • Supports 3Rs - reduces animal cardiac studies
  • Dose-dependent risk assessment - simulations across a full concentration range.

In silico simulations can predict TdP risk

In silico simulations capture critical multi-channel interactions that single-channel screening cannot reveal. Balanced ion channel‐blocking of ICaL and INaL can offset hERG-mediated reductions in repolarizing current, reducing proarrhythmic risk despite IKr block. In other words, blocking one channel can counteract the effects of blocking another and reduce TtP risk. This mechanistic, multi-channel approach integrated with in silico cardiac models provides improved TdP prediction compared to hERG assessment alone.
For example, Verapamil, which blocks calcium ion channels, has low TdP risk despite also blocking hERG channels (Figure 2). By simulating how compounds affect the balance of inward and outward ionic currents, in silico models provide mechanistic insight into why some hERG blockers remain clinically safe while others cause dangerous arrhythmias.

Early assessment and comparison to established compounds

ApconiX in silico reports compare client compounds to a range of drugs with established risk profiles at their clinical Cmax (Figure 2 &3), enabling detection of proarrhythmic liabilities early. Representative traces of the simulated cardiac action potentials are provided for all reference drugs and client compounds (Figure 2). Metrics that predict a compounds TdP risk, including qNet and APD90 (action potential duration), are also provided through analysis of the simulated action potentials (Figure 3). 

In silico models enable detection of proarrhythmic liabilities early in development. Additionally, in silico modelling provides a concentration-dependent risk profile across a wide range of exposures, allowing programmes to understand not just whether a compound poses a level risk, but at what concentration that risk becomes relevant. 

insilico tdp risk stratification | ApconiX

Figure 3. Metrics for TdP risk stratification.

(Top) APD90 and qNet values for the 12 CiPA training compounds at their clinical Cmax, grouped by TdP risk category. Horizontal dashed lines indicate control (no drug) values. qNet shows clearer separation between risk categories than APD90. (Bottom) Concentration-response curves for test Compounds A and B demonstrate progressive APD90 prolongation and qNet reduction, indicating increased TdP risk at higher concentrations. Dashed horizontal lines indicate control (black).

in silico referance compounds | ApconiX

Figure 2. In silico action potential simulations. 

Simulated cardiac action potentials using the O’Hara-Rudy model at Cmax (reference compounds) or across concentration ranges (test compounds). Reference compounds from the CiPA training set are grouped by TdP risk: low-risk drugs (green) often show minimal APD prolongation, intermediate-risk drugs (yellow-orange) show moderate prolongation, and high-risk drugs (red) show severe prolongation with delayed repolarization. Test Compounds A and B (blue gradient, 0-10 µM) demonstrate concentration-dependent APD prolongation. Scale bars: 20 mV (vertical), 100 ms (horizontal). Control (no drug) shown in black.

In Silico Cardiac Safety Modelling FAQs

There is no minimum stage requirement. Earlier screening helps de-risk compounds before costly in vivo or clinical studies.  

Yes. In silico modelling can reveal whether hERG block is proarrhythmic at the compounds Cmax and whether activity on other channels (e.g. calcium or late sodium) provides a counterbalancing effect.  

A written report comparing your compound(s) to reference drugs with known clinical TdP risk, including simulated action potential traces, APD₉₀ and qNet metrics. Each report also includes a detailed results text contextualising the findings. 

In silico modelling provides mechanistic insight from ion channel data and can help prioritise which compounds warrant more expensive wet-lab follow-up. 

The 2022 ICH E14/S7B Q&A document recognises in silico models as an optional follow-up tool for TdP risk characterisation.  

Yes. In silico modelling can reduce the need for animal cardiac studies by providing mechanistic safety data in silico, which is recognised under 3R principles. 

No. In silico modelling sits between ion channel screening and iPSC-cardiomyocyte assays. Human iPSC-derived cardiomyocytes (hiPSC-CMs) provide a more physiologically complete cellular model, capable of detecting missed or unanticipated electrophysiological responses not identified by ionic current or in silico models alone, including effects on contractility.  

In silico modelling requires no wet-lab work and can typically be turned around rapidly once IC₅₀ data is supplied. 

Compounds can be evaluated across broad concentration ranges without known Cmax. Compounds with Cmax data will receive an additional metric, the Torsade Metric Score (TMS). TMS is defined as the average qNet value at 1-4x Cmax which has been shown to be a better predictor compared to APD or qNet.  

Minimum: hERG IC50.  

Optimal: Full CiPA panel (IKr, INa, INaL, ICaL, IKs, Ito, IK1) for maximum insight. This is recommended.  

Yes. 

Disclaimer

In silico cardiac modelling provides mechanistic risk assessment to complement in vitro and clinical data. Predictions should be interpreted by qualified scientists within the context of integrated drug development. This service is for research use; clinical decision-making requires comprehensive safety evaluation across multiple data types.

Want to find out more?

If you have ion channel screening data, ApconiX can provide an in silico cardiac safety report quickly.
Get in touch to discuss how this fits into your programmes.