Applications of machine learning in decision analysis for dose management for dofetilide.

BACKGROUND: Initiation of the antiarrhythmic medication dofetilide requires an FDA-mandated 3 days of telemetry monitoring due to heightened risk of toxicity within this time period. Although a recommended dose management algorithm for dofetilide exists, there is a range of real-world approaches to dosing the medication. METHODS AND RESULTS: In this multicenter investigation, clinical data from the Antiarrhythmic Drug Genetic (AADGEN) study was examined for 354 patients undergoing dofetilide initiation. Univariate logistic regression identified a starting dofetilide dose of 500 mcg (OR 5.0, 95%CI 2.5-10.0, p<0.001) and sinus rhythm at the start of dofetilide loading (OR 2.8, 95%CI 1.8-4.2, p<0.001) as strong positive predictors of successful loading. Any dose-adjustment during loading (OR 0.19, 95%CI 0.12-0.31, p<0.001) and a history coronary artery disease (OR 0.33, 95%CI 0.19-0.59, p<0.001) were strong negative predictors of successful dofetilide loading. Based on the observation that any dose adjustment was a significant negative predictor of successful initiation, we applied multiple supervised approaches to attempt to predict the dose adjustment decision, but none of these approaches identified dose adjustments better than a probabilistic guess. Principal component analysis and cluster analysis identified 8 clusters as a reasonable data reduction method. These 8 clusters were then used to define patient states in a tabular reinforcement learning model trained on 80% of dosing decisions. Testing of this model on the remaining 20% of dosing decisions revealed good accuracy of the reinforcement learning model, with only 16/410 (3.9%) instances of disagreement. CONCLUSIONS: Dose adjustments are a strong determinant of whether patients are able to successfully initiate dofetilide. A reinforcement learning algorithm informed by unsupervised learning was able to predict dosing decisions with 96.1% accuracy. Future studies will apply this algorithm prospectively as a data-driven decision aid.

Assessment of conventional cardiovascular risk factors and multiple biomarkers for the prediction of incident heart failure and atrial fibrillation.

OBJECTIVES: the purpose of this study was to assess the predictive accuracy of conventional cardiovascular risk factors for incident heart failure and atrial fibrillation, and the added benefit of multiple biomarkers reflecting diverse pathophysiological pathways. BACKGROUND: heart failure and atrial fibrillation are interrelated cardiac diseases associated with substantial morbidity and mortality and increasing incidence. Data on prediction and prevention of these diseases in healthy individuals are limited. METHODS: in 5,187 individuals from the community-based MDCS (Malmö Diet and Cancer Study), we studied the performance of conventional risk factors and 6 biomarkers including midregional pro-atrial natriuretic peptide (MR-proANP), N-terminal pro-B-type natriuretic peptide (NT-proBNP), midregional pro-adrenomedullin, cystatin C, C-reactive protein (CRP), and copeptin. RESULTS: during a mean follow-up of 14 years, 112 individuals were diagnosed with heart failure and 284 individuals with atrial fibrillation. NT-proBNP (hazard ratio [HR]: 1.63 per SD, 95% confidence interval [CI]: 1.29 to 2.06, p < 0.001), CRP (HR: 1.57 per SD, 95% CI: 1.28 to 1.94, p < 0.001), and MR-proANP (HR: 1.26 per SD, 95% CI: 1.02 to 1.56, p = 0.03) predicted incident heart failure independently of conventional risk factors and other biomarkers. MR-proANP (HR: 1.62, 95% CI: 1.42 to 1.84, p < 0.001) and CRP (HR: 1.18, 95% CI: 1.03 to 1.34, p = 0.01) independently predicted atrial fibrillation. Addition of biomarkers to conventional risk factors improved c-statistics from 0.815 to 0.842 for heart failure and from 0.732 to 0.753 for atrial fibrillation and the integrated discrimination improvement for both diseases (p < 0.001). Net reclassification improvement (NRI) with biomarkers was observed in 22% of individuals for heart failure (NRI, p < 0.001) and in 7% for atrial fibrillation (NRI, p = 0.06), mainly due to up-classification of individuals who developed disease (heart failure: 29%, atrial fibrillation: 19%). Addition of CRP to natriuretic peptides did not improve discrimination or reclassification. CONCLUSIONS: conventional cardiovascular risk factors predict incident heart failure and atrial fibrillation with reasonable accuracy in middle-age individuals free from disease. Natriuretic peptides, but not other biomarkers, improve discrimination modestly for both diseases above and beyond conventional risk factors and substantially improve risk classification for heart failure.

Electrocardiographic assessment for therapeutic proteins--scientific discussion.

Electrocardiographic monitoring is an integral component of the clinical assessment of cardiac safety of all compounds in development. The International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use E14 guideline recommends a dedicated study to evaluate drug-induced effects on cardiac repolarization ("thorough QT/QTc study"). There has been limited published information on QT interval changes secondary to therapeutic proteins; however, in theory, biologic therapies may affect cardiac electrical activity either directly or indirectly. This article summarizes scientific discussions of members of the Cardiac Safety Research Consortium and includes possible approaches to consider for the clinical evaluation of drug-induced QT prolongation in development programs of therapeutic proteins.

Evaluation of ventricular arrhythmias in early clinical pharmacology trials and potential consequences for later development.

This white paper, prepared by members of the Cardiac Safety Research Consortium, discusses several important issues regarding the evaluation of ventricular arrhythmias in early clinical pharmacology trials and their potential consequences for later clinical drug development. Ventricular arrhythmias are infrequent but potentially important medical events whose occurrence in early clinical pharmacology trials can dramatically increase safety concerns. Given the increasing concern with all potential safety signals and the resultant more extensive electrocardiographic monitoring of subjects participating in early phase trials, an important question must be addressed: Are relatively more frequent observations of ventricular arrhythmias related simply to more extensive monitoring, or are they genuinely related to the drug under development? The discussions in this paper provide current thinking and suggestions for addressing this question.