Inhibition of the human Ether-a-go-go-Related Gene (hERG) potassium channel by medications appears to be the most common mechanism of acquired QT interval prolongation.

Why is hERG Affinity Important?
QT interval prolongation is a side effect induced by structurally diverse drugs that has been linked to life threatening ventricular arrhythmias including Toursade de Pointes. Because more and more non-antiarrhythmic drugs are being shown to have the potential to prolong QT interval, it is important that all new chemical entities (NCE) are thoroughly investigated for this potential early in their preclinical development. Therefore, in silico prediction of the hERG screening plays an important role in understanding the hERG-drug binding. Such predictive hERG models are highly valuable as in vitro and in vivo measurements are costly, labour intensive and not widely available. In recent years, structure-activity studies on the growing number of marketed drugs and investigational compounds exhibiting inadvertent hERG channel blockade have been reported.

hERG pIC50 Model
Data on hERG K+ channel blockers were derived from various literature sources. 168 structures with patch-clamp IC50 values for inhibition of hERG K+ channels expressed in mammalian cells were selected, as this is the 'gold-standard' experimental technique for determining hERG inhibition. Other higher-throughput approaches to measuring hERG inhibition show poor correlation with patch-clamp measurements in mammalian cells and with each other and were therefore not used in the development of this model. A model was built using the non-linear Gaussian Processes technique implemented within the StarDrop Auto-Modeller. The model was trained on 135 compounds and tested on 33 compounds. The R2 value for the training set is 0.78 and the RMSE is 0.66 log units. On test set (33 compounds) the model achieved an R2 value of 0.72 (R2corr=0.74) and a RMSE in prediction of 0.64 log units.