Scholarlink

<p>Torsade de Pointes (TdP) is a potentially fatal ventricular arrhythmia associated with drug-induced disturbances of cardiac repolarization. Because QT interval prolongation alone provides limited mechanistic specificity, the Comprehensive in vitro Proarrhythmia Assay (CiPA) paradigm promotes integrated evaluation using ion-channel pharmacology, <i>in silico</i> electrophysiology, and human stem-cell–based evidence. In this study, we benchmarked the Paci2020 human induced pluripotent stem cell–derived cardiomyocyte (hiPSC-CM) <i>in silico</i> model for CiPA-aligned TdP risk prediction using an interpretable ordinal logistic regression (OLR) framework. Ten mechanistically grounded biomarkers spanning ionic charge balance, action-potential morphology, and calcium handling dynamics were extracted from simulations of 28 CiPA reference drugs across clinically relevant exposures (up to fourfold free peak plasma concentration, Cmax) and evaluated using CiPA-recommended diagnostic metrics under repeated resampling. Single-biomarker models—including qNet and APD90 (a net ionic charge metric and action potential duration at 90% repolarization)—exhibited strong discrimination in some comparisons but did not consistently meet CiPA confidence requirements for reliably excluding high-risk compounds, which motivated the integration of multiple biomarkers. Exhaustive screening of all non-empty biomarker subsets (1023 combinations) identified a parsimonious four-biomarker signature (qNet, Ca_resting, dVm/dtmax, Cycle_length) that satisfied all CiPA acceptance criteria on the independent test set, outperforming the Tomek-ORd benchmark and approaching the performance of the CiPAORdv1.0 adult-ventricular model. Collectively, these findings demonstrate that, despite the immature electrophysiology of hiPSC-CMs, CiPA-grade TdP risk prediction can be achieved through systematic, mechanistically interpretable multi-biomarker integration, supporting the role of hiPSC-CM <i>in silico</i> models as complementary tools for early-stage cardiotoxicity screening.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

In silico cardiac safety assessment using a multi-biomarker approach based on an electrophysiological model of hiPSC-derived cardiomyocytes

  • 期刊论文

  • 发表日期：

    2026年3月19日

查看全文
  • Ari Kurniawan Saputra,
  • Ali Ikhsanul Qauli,
  • Aroli Marcellinus,
  • Ki Moo Lim

摘要

Torsade de Pointes (TdP) is a potentially fatal ventricular arrhythmia associated with drug-induced disturbances of cardiac repolarization. Because QT interval prolongation alone provides limited mechanistic specificity, the Comprehensive in vitro Proarrhythmia Assay (CiPA) paradigm promotes integrated evaluation using ion-channel pharmacology, in silico electrophysiology, and human stem-cell–based evidence. In this study, we benchmarked the Paci2020 human induced pluripotent stem cell–derived cardiomyocyte (hiPSC-CM) in silico model for CiPA-aligned TdP risk prediction using an interpretable ordinal logistic regression (OLR) framework. Ten mechanistically grounded biomarkers spanning ionic charge balance, action-potential morphology, and calcium handling dynamics were extracted from simulations of 28 CiPA reference drugs across clinically relevant exposures (up to fourfold free peak plasma concentration, Cmax) and evaluated using CiPA-recommended diagnostic metrics under repeated resampling. Single-biomarker models—including qNet and APD90 (a net ionic charge metric and action potential duration at 90% repolarization)—exhibited strong discrimination in some comparisons but did not consistently meet CiPA confidence requirements for reliably excluding high-risk compounds, which motivated the integration of multiple biomarkers. Exhaustive screening of all non-empty biomarker subsets (1023 combinations) identified a parsimonious four-biomarker signature (qNet, Ca_resting, dVm/dtmax, Cycle_length) that satisfied all CiPA acceptance criteria on the independent test set, outperforming the Tomek-ORd benchmark and approaching the performance of the CiPAORdv1.0 adult-ventricular model. Collectively, these findings demonstrate that, despite the immature electrophysiology of hiPSC-CMs, CiPA-grade TdP risk prediction can be achieved through systematic, mechanistically interpretable multi-biomarker integration, supporting the role of hiPSC-CM in silico models as complementary tools for early-stage cardiotoxicity screening.