Abstract Body: Background: Thorough QT studies using Electrocardiogram data, are a well-established method for testing the pro-arrhythmic propensity of drugs performed during drug development. ECG analysis is currently reduced to simplified biomarkers, such as rate, intervals and amplitudes. However, ECG waveforms are complex and generally sampled with high fidelity 125-1000 Hz. Our novel mathematical method, Symmetric Projection Attractor Reconstruction analyses the morphology and variability of such waveforms, using every data point to generate a new, faithful 2D waveform representation termed an attractor. Moxifloxacin know QT prolonging medicine and has been shown to cause a mean increase of the QTc interval of 10–14 mS after a single dose of 400 mg. Separately, studies have shown that meals can alter QT interval as well as altering the pharmacokinetic profile of moxifloxacin itself with gender and ethnicity impacting individual responses. In-depth knowledge of these physiological factors could support safer medicines development and clinical use.
Purpose: This project investigated the feasibility of SPAR analysis of ECG signals to detect drug induced changes which a higher degree of sensitivity, to support safety evaluation during medicines development
Methods: High resolution triplicate 12 lead ECG time-series data from a Phase one study of healthy human volunteers cross over of placebo, moxifloxacin or moxifloxacin + food was retrospectively analysed using SPAR at 0,1 and 6 hours. MATLAB bespoke software tool was used to perform SPAR analyses to transform ECG time-series into corresponding attractors. Corresponding single point measurements were previously obtained as part of the original Phase one study.
Results: Initial evaluation of attractors revealed high inter-individual variation, consistent with previous SPAR ECG analysis. Nevertheless, we identified changes in subjects receiving Moxifloxacin +/- food vs placebo control. These changes were not as evident in previously obtained QT intervals.
Conclusion: This pilot study illustrates that SPAR is highlighting changes on the ECG in fasted and fed states that are less obvious from QT data alone. Further evaluation of the whole dataset to identify generalisable SPAR features, and their comparison with existing markers is now necessary. The individualised nature of ECGs and corresponding attractors indicates the method may have highest utility in personalised medicine.