An oxygen-permeable, stirrer-pump-integrated liver-heart microphysiological system enables time-resolved analysis in PK-PD relationships of metabolism-dependent cardiotoxicity.

Predicting metabolism-dependent drug toxicity encompassing both the bioactivation and detoxification processes remains a critical challenge in preclinical safety evaluations. Conventional cardiomyocyte-only assays frequently misclassify compounds such as terfenadine, which require hepatic metabolism to attenuate cardiotoxicity. Current microphysiological systems (MPS) have practical limitations, such as insufficient hepatocyte oxygenation and complex external fluidic setups. We developed a liver-heart co-culture system that integrates three-dimensional human iPS cell-derived engineered heart tissues (hiPSC-EHTs) and cryopreserved primary human hepatocytes (PHHs), using a modified BioStellar™ Plate-a tubeless, stirrer-pump-integrated microfluidic device customized with an oxygen-permeable membrane. Using this system, we evaluated the cardiotoxicity of terfenadine in hiPSC-EHT monocultures and liver-heart co-cultures. Terfenadine exposure induced marked contractile dysfunction in hiPSC-EHT monocultures, whereas co-culturing with PHHs significantly attenuated this effect. Functional monitoring revealed time- and dose-dependent recovery, consistent with a largely reversible pharmacological mechanism. Drug concentration analysis by LC-MS/MS showed a marked increase in the formation of the primary metabolite fexofenadine, in the co-culture system, which was consistent with the reduced cardiotoxic potential of this metabolite. Gene expression analysis confirmed the presence of hepatic CYP3A4, which supports its predominant role in metabolic detoxification. This study demonstrated that a tubeless, stirrer-pump-integrated system with sustained hepatocyte function enables time-resolved evaluation of pharmacokinetic-pharmacodynamic relationships of metabolism-dependent cardiotoxicity on a single platform. This proof-of-concept approach suggests a potential improvement in the translational predictability of human-relevant, metabolism-dependent cardiotoxicity by enabling integrated evaluation of pharmacokinetic and pharmacodynamic responses. These findings highlight the utility of this system as a simple and scalable alternative to conventional perfusion-based MPS.