Microfluidic liver-chip to predict clinical drug-drug interaction net effects for precipitant drugs with dual CYP3A4 inhibition and induction potential.

Accurate predictions of complex clinical drug-drug interactions (DDIs), arising from dual induction and time-dependent inhibition (TDI) of CYP3A4, has remained challenging with conventional in vitro and static/dynamic modeling approaches. In this work, we aimed to anticipate the hepatic DDI effects of 6 CYP3A4 precipitant drugs using a liver-chip coupled to microfluidic perfusion, which enabled simulating clinically relevant pharmacokinetic (PK) time-concentration profiles. Midazolam clearance was measured in the liver-chip to determine the CYP3A4-mediated DDI net effect following exposure to precipitants under either dynamic or constant concentration conditions. For direct comparison, hepatic DDI reference values were generated based on clinical DDI studies by physiologically based PK modeling. Under microfluidic perfusion, CYP3A4 activity in the liver-chip was retained for 3 days and inducible by rifampicin. Although most precipitant drugs induced CYP3A4 mRNA levels, CYP3A4 activity net effects showed either induction or inhibition, in line with clinical observations. Compared with a mechanistic static net effect model and physiologically based PK simulations, liver-chip predictions showed closer alignment and higher accuracy relative to hepatic reference values. Although constant exposures offered the strongest quantitative performance, the ability to apply dynamic PK profiles represents a distinctive feature of this platform. Collectively, these findings position the human liver-chip as a novel translational platform for predicting complex hepatic CYP3A4 DDIs arising from dual TDI and induction in a single, holistic in vitro model. This approach aligns with the broader global regulatory shift toward human-relevant new approach methodologies. SIGNIFICANCE STATEMENT: Complex CYP3A4-mediated drug-drug interactions (DDIs), driven by concurrent time-dependent inhibition and induction, remain difficult to anticipate using conventional in vitro and modeling frameworks. This study establishes a human liver-chip model capable of capturing dual CYP3A4 inhibition and induction within a single experiment, and evaluates how clinical exposure regimes shape the quantitative prediction of hepatic DDI effects. Relative to modeling approaches, this platform demonstrates superior accuracy against hepatic DDI references, supporting more translational DDI risk assessments.