Summary
Objectives:
Estimation of possible cardiovascular side effects belongs to the safety assessment
of every drug candidate. Drug-induced prolongation of the QT interval can result in
life-threatening ventricular arrhythmia. In pre-clinical drug development, animal
experiments are used to study this possible effect. Researchers have become aware
that correction formulae derived for human beings are not applicable to animal experiments.
Methods:
We investigated some of the proposed models by comparing the outcomes of the analyses
on the same data. The data was derived from telemetry measurements on Labrador dogs.
We propose the use of both the correlation with heart rate (or RR interval) and a
measure of predictive performance. As a sufficiently large number of observations
were available, the data was subdivided into a training and a test set. The training
set serves to estimate the respective parameters while the test set is used to determine
the performance of the model. Here, a kind of PRESS statistic was used. Next, the
models were considered for treated animals, using the estimated parameters. Both positive
and negative controls were used.
Conclusions:
Most models under consideration performed quite well. These models eliminated the
correlation for the most part and were reasonably predictive. Furthermore, they reliably
differentiate between positive and negative controls. The next steps in identifying
the best correction will be to consider additional compounds as well as other species
to validate our current results.
Keywords
QT prolongation - correction formulae - comparison - prediction
