Create new account

Cardiotoxicity in vitro


Torsades de Pointes (TdP) is defined as polymorphic ventricular tachycardia with a characteristic twist of the QRS complex around the isoelectric baseline on the ECG. The most common cause of TdP is long QT syndrome (LQTS) either inherited or acquired. Most, if not all drugs which induce QT interval prolongation have been reported to cause this through the inhibition of the rapid component of the delayed repolarizing current (IKr) due to hERG channel blockade. Although TdP occurrence in humans is extremely rare, as it can degenerate into potentially fatal ventricular fibrillation TdP constitutes unacceptable risk especially for drug developed for non-life-threatening condition treatment. In recent years drug-induced arrhythmias are of considerable concern during drug discovery and development process as the assessment of torsadogenic risk is required by the regulatory agencies during the registration process. A number of drugs were withdrawn from market and prescription of many others has been significantly restricted due to an unexpected occurrence of cardiotoxic effects during their application.

Current cardiac safety assessments during R&D process relay of “surrogate markers” of TdP. According to ICH guideline hERG channel function studies are one of the options among the battery of non-clinical tests used to cardiac liability evaluation and the half maximal inhibitory concentration (IC50) of a compound can serve as TdP risk marker and in the industry is a basis for go-no go decisions making at early development stages.

Traditional in vitro approach to drug-hERG electrophysiological interaction assessment as Patch Clamp technique which is currently accepted ‘gold standard’ is time-, labour- and cost-consuming. Moreover a low throughput of these assays impedes their application for compounds’ cardiac liability screening at early stages of drug development. Thus there is a considerable demand for accurate, cheap and fast computational toxicology tools which can accelerate cardiotoxic potential assessment through in silico modeling.

The main hERG project goals are development of the data base relevant for torsadogenic risk evaluation, inter-system extrapolating factors for experimental IC50 values and building predictive models for cardiotoxic liability of compounds.


For the performance of every computational model quality of the data used for its development is crucial issue. Unfortunately, there is limited public availability of good-quality toxicity data. For that reasons the first step of hERG project was the hERG IC50 data collection and assessment in order to provide high-quality data set for further studies. Over 300 papers regarding IKr inhibition experimental evaluations were found in Medline, Scopus and Google Scholar databases. In addition to half-maximal inhibitory concentration (IC50 value), papers were revised for additional information as follows:

  • model applied in experiments

  • measurement technique

  • type of transfection

  • experimental conditions (temperature and K+ concentration in external solution)

  • voltage protocol (protocol type: step, ramp, step-ramp; holding potential, depolarization potential, measurement potential; depolarization pulse time)

As the vast majority of IC50 values gathered were obtained using either XO (Xenopus oocytes) or HEK (human embryonic kidney) or CHO (Chinese hamster ovary) expression system the initial data base was restricted to records regarding these three experimental models. Final data set consists of 601 IC50 values and the complete set of information was collected for 500 of them. Comprehensive data base description as well as basic data analysis can be found in our publication. The complete data files (either .xlsx or .ods) can be also downloaded directly from the repository (PLEASE NOTE - available after registration!).

Extrapolating factors

The hERG activity of the compound characterizes its pro-arrhythmic potential. Experience gained during drug development processes in pharma industry indicates that compound with the IC50 value above the certain threshold in the in vitro tests can be problematic at later stages. However, drawing reliable conclusions and correct decision making on tests continuation or compound withdrawal require all IC50 values to be obtained in identical experimental settings. Literature data analysis shows that the hERG interactions experiments carried in different conditions with use of different in vitro systems for the same substance can results with different IC50 values. Therefore the second component of the hERG project is a set of extrapolation factors enabling the more flexible choice of some elements of the experimental procedure without results significance depreciation. In our believe proposed inter-system and inter-temperature extrapolating factors can be useful tools to support cardiac liability evaluation, based on hERG IC50 values derived from miscellaneous in vitro systems. Factors development and evaluation process with conclusions was presented in paper accepted for publication in Toxicology Mechanisms and Methods.


Computational assays do not require any special equipment nor reagents and synthesized molecules are necessary. The value of in sillico modeling tools arises especially from their applicability early in development.

We are currently working on in silico drug-hERG interaction models development. One of the main goals of hERG project is to create a qualitative virtual screen model and a quantitative model for predicting hERG inhibition.

Different statistical and computational approaches are tested:

  • non-linear statistical models

  • artificial neural networks

  • decision trees and forests

  • Bayes algorithms

  • rough and fuzzy sets


  • RBF

  • kNN

  • other

Developed models will be validated with diverse statistical accuracy measures. Generalization ability of the models will be also assessed on the test set of newly investigated compounds which is currently under development.