A new classifier-based strategy for in-silico ion-channel cardiac drug safety assessment

Abstract

There is currently a strong interest in using high-throughput in-vitro ion-channel screening data to make predictions regarding the cardiac toxicity potential of a new compound in both animal and human studies. A recent FDA think tank encourages the use of biophysical mathematical models of cardiac myocytes for this prediction task. However, it remains unclear whether this approach is the most appropriate. Here we examine five literature data-sets that have been used to support the use of four different biophysical models and one statistical model for predicting cardiac toxicity in numerous species using various endpoints. We propose a simple model that represents the balance between repolarisation and depolarisation forces and compare the predictive power of the model against the original results (leave-one-out cross-validation). Our model showed equivalent performance when compared to the four biophysical models and one statistical model. We therefore conclude that this approach should be further investigated in the context of early cardiac safety screening when in-vitro potency data is generated.

In Silico Pharmacoepidemiologic Evaluation of Drug-Induced Cardiovascular Complications Using Combined Classifiers

“…15 Building on this, Hitesh and co-workers built classifiers for assessment of drug cardiotoxicity with accuracies ranging from 0.675 to 0.95 by leave-one-out cross validation. 16 Reported studies thus far are largely limited by use of only a single machine-learning algorithm with low or moderate accuracy. In order to advance the field of drug development, it is vital to develop robust and effective in silico models with high accuracy for evaluation of drug-induced cardiotoxicity.…”

Intrinsic structure of model-derived metrics for in silico proarrhytmic risk assessment identified by global sensitivity analysis

“…We have recently shown comparable accuracies for TdP risk classification from direct 476 features in tests against several of the other published methods [14]. Few previous studies [13,15,16] have also reported 477 similar findings for particular datasets where metrics based on direct features provided high predictive power. probability to obtain one that performs as well as qN et.…”

“…The second group of coefficient sets (group B) was more inclusive and allowed the same admissible range ([-1, 1]) 305 for all coefficients. All the linear combination metrics generated by(13) according to the coefficients in group A and 306 group B were then used in multiclass logistic regression for tertiary classification of the 12 CiPA training drugs. Out of 307 the two groups of random sets of coefficients, approximately 1% (n=400) and 0.25% (n=100) of the metrics generated 308 from group A and B, respectively, resulted in perfect discrimination of the drugs into low, intermediate, and high risk 309…”

Mechanistic Systems Modeling to Improve Understanding and Prediction of Cardiotoxicity Caused by Targeted Cancer Therapeutics

“…In this context, models that describe biological mechanisms through differential equations are frequently referred to as QSP models (Leil and Bertz, 2014;Gadkar et al, 2016). Although precise definitions remain a matter of debate, QSP models are generally distinguished from both purely empirical, statistical approaches such as computing a risk score for a drug based on a series of measurements Mistry et al, 2015), and pharmacokinetic models that can predict the effects of dosing on cardiotoxicity (van Hasselt et al, 2012) but generally offer only limited mechanistic insight. Although QSP models have been exploited to understand cardiotoxicity caused by anthracyclines (de Oliveira et al, 2016), the application of QSP to TKI-induced cardiotoxicity is still in its early stages.…”