中文导读
ADMET 性质决定候选药物的成败。本文介绍器官芯片在吸收、分布、代谢、排泄与毒性研究中的应用与优势。
以下为英文全文（English full text）
1. Introduction
Absorption, Distribution, Metabolism, Excretion, and Toxicity (ADMET) properties are fundamental determinants of drug efficacy and safety. Poor ADMET profiles are responsible for approximately 50% of drug failures in clinical development, making early and accurate prediction of these properties a critical priority in pharmaceutical research. Organ-on-chip (OoC) microphysiological systems offer unprecedented capabilities for integrated ADMET assessment by recapitulating human tissue architecture, physiological fluid flow, and organ-organ communication in a controlled in vitro environment. This article provides a comprehensive technical overview of OoC-based ADMET studies, including specific protocols, analytical parameters, and validation strategies.
2. The ADMET Challenge in Drug Development
2.1 Traditional ADMET Assay Limitations
Conventional ADMET assessment relies on a patchwork of disparate assays: Caco-2 monolayers for intestinal permeability, liver microsomes or hepatocyte suspensions for metabolism, transporter-overexpressing cell lines for distribution, and immortalized renal cell lines for excretion studies. While these assays provide useful data, they fail to capture the integrated physiology of the human body. Static 2D cultures lack tissue architecture, fluid flow, and organ-organ communication, leading to poor prediction of systemic pharmacokinetics (PK) and pharmacodynamics (PD). Inter-species differences in drug metabolism (e.g., cytochrome P450 enzyme expression) further limit the translatability of animal-based ADMET data.
2.2 The Organ-on-Chip Advantage
OoC systems address these limitations by integrating multiple tissue models within physiologically relevant microenvironments. Shroff et al. (2022) reviewed multi-organ chips for metabolism studies, highlighting that these systems enable real-time assessment of organ cross-talk during drug exposure. Unlike static cultures, OoC platforms maintain tissue-specific gene expression, metabolic activity, and barrier function over extended culture periods (weeks to months), enabling chronic ADMET studies that were previously impossible in vitro.
3. Absorption Studies on Organ-on-Chip Platforms
3.1 Intestinal Absorption Models
The human intestine is the primary site of oral drug absorption, where compounds encounter complex epithelial barriers, mucus layers, microbiome interactions, and active transport mechanisms. Gut-on-chip devices replicate these features using porous membranes seeded with intestinal epithelial cells (Caco-2, HT29-MTX, or primary enterocytes) on the apical side and microvascular endothelial cells on the basolateral side. Cyclical mechanical strain (5–10%) mimics peristaltic motion, enhancing epithelial differentiation and mucus production.
For permeability assessment, drugs are introduced into the apical (luminal) compartment, and apparent permeability (Papp) is calculated from basolateral concentrations measured over time: Papp = (dQ/dt) / (A × C0), where dQ/dt is the transport rate, A is the membrane area, and C0 is the initial donor concentration. Transwell-compatible OoC platforms enable direct comparison with gold-standard Caco-2 assays, with typical values of Papp < 1 × 10⁻⁶ cm/s indicating low permeability, 1–10 × 10⁻⁶ cm/s moderate, and >10 × 10⁻⁶ cm/s high permeability.
3.2 P-glycoprotein and Transporter Interactions
Active efflux by P-glycoprotein (P-gp/ABCB1) and other ABC transporters significantly limits intestinal absorption of many drugs. OoC platforms enable bidirectional transport studies (A-to-B and B-to-A) to calculate efflux ratios (ER = Papp(B-A) / Papp(A-B)). ER > 2 indicates active efflux and potential P-gp substrate liability. Verapamil (100 µM) and cyclosporine A (10 µM) serve as positive control inhibitors, reducing efflux ratios by >50% in validated systems.
4. Distribution Studies: Blood-Brain Barrier and Tissue Penetration
4.1 Blood-Brain Barrier (BBB) on Chip
The BBB restricts drug access to the central nervous system (CNS) through tight junctions between brain microvascular endothelial cells, pericytes, and astrocyte end-feet. BBB-on-chip devices replicate this neurovascular unit by co-culturing endothelial cells (hBMECs or iPSC-derived) with pericytes and astrocytes on opposite sides of a porous membrane under physiological shear stress (0.1–5 dyne/cm²). Trans-endothelial electrical resistance (TEER) values >200 ohm·cm² indicate robust barrier formation.
Apparent permeability across the BBB is measured using fluorescently labeled compounds (e.g., sodium fluorescein, MW 376 Da) or radiolabeled drugs. The permeability-surface area product (PS) is calculated as: PS = -ln(1 - Cr/Cd) / (1/VD + 1/VR), where Cr and Cd are receiver and donor concentrations, and VD and VR are donor and receiver volumes. Compounds with PS < 1 × 10⁻⁶ mL/s/g are considered to have poor CNS penetration.
4.2 Tissue Distribution via Multi-Organ Chips
Systemic distribution is modeled using multi-organ chips connected by a shared vascular circuit. Shuler and colleagues demonstrated that a body-on-a-chip integrating gut, liver, and target tissues could predict drug distribution to organs based on perfusion rates, plasma protein binding, and tissue partition coefficients. By sampling media from each organ compartment at defined intervals, researchers generate concentration-time profiles analogous to clinical PK studies.
5. Metabolism Studies: Liver and Extra-Hepatic Metabolism
5.1 Liver-on-Chip for Drug Metabolism
The liver is the primary site of drug metabolism, and liver-on-chip platforms have emerged as the most physiologically relevant in vitro models for metabolic stability and metabolite identification. HepaRG cells, primary human hepatocytes, or iPSC-derived hepatocytes are cultured under perfusion (0.5–2 µL/min) with supporting endothelial and stellate cell populations. Key metabolic parameters include:
- CYP enzyme activity: Measured using specific substrates (phenacetin for CYP1A2, diclofenac for CYP2C9, bufuralol for CYP2D6, midazolam for CYP3A4)
- Intrinsic clearance (CLint): Calculated from first-order depletion kinetics: CLint = (ln(C0/Ct)) / (t × cell number), normalized to 10⁶ cells
- Metabolite identification: LC-MS/MS analysis of culture media and cell lysates for phase I (oxidation) and phase II (conjugation) metabolites
Liver-on-chip platforms maintain CYP activity for >4 weeks, compared to <1 week in static hepatocyte cultures, enabling chronic drug metabolism and enzyme induction studies. Rifampicin (10 µM, 72 hours) serves as a positive control for CYP3A4 induction, typically increasing activity 3–5 fold.
5.2 Gut-Liver Axis for First-Pass Metabolism
The gut-liver chip integrates intestinal absorption with hepatic first-pass metabolism, providing a complete model of oral bioavailability. Shuler et al. demonstrated that this co-culture system successfully matched 22 of 24 clinically relevant oral bioavailability predictions, outperforming isolated Caco-2 or hepatocyte models. In this configuration, compounds absorbed across the intestinal epithelium are transported directly to the liver compartment via physiological flow, where metabolic transformation occurs before systemic distribution.
6. Excretion Studies: Kidney-on-Chip for Renal Clearance
6.1 Renal Clearance Mechanisms
The kidney eliminates drugs and metabolites through glomerular filtration, tubular secretion, and reabsorption. Kidney-on-chip devices use proximal tubule epithelial cells (RPTEC/TERT1, iPSC-derived, or primary) cultured on porous membranes under physiological shear stress (0.1–2 dyne/cm²) and transepithelial pressure gradients. TEER values of 100–400 ohm·cm² confirm tight junction formation and barrier integrity.
Clearance is measured by introducing drugs into the vascular (basolateral) compartment and monitoring their appearance in the urinary (apical) compartment. Renal clearance (Clr) is calculated as: Clr = (Ur × Vr) / (Cp × t), where Ur is urinary concentration, Vr is urine volume, Cp is plasma concentration, and t is collection time. Active secretion is assessed using probe substrates for specific transporters (p-aminohippurate for OAT1/3, cimetidine for OCT2, digoxin for P-gp).
6.2 Drug-Drug Interactions at the Kidney
Transporter-mediated drug-drug interactions (DDIs) are a major cause of altered renal excretion. OoC platforms enable mechanistic studies of competitive inhibition at renal transporters. For example, probenecid (1 mM) inhibits OAT1-mediated secretion of cephalosporins, increasing their plasma half-life. These studies are critical for predicting clinical DDI risk and optimizing dosing regimens.
7. Toxicity Assessment in Organ-on-Chip Systems
7.1 Hepatotoxicity on Chip
Drug-induced liver injury (DILI) is a leading cause of drug withdrawal and clinical trial failure. Liver-on-chip platforms enable early detection of hepatotoxicity through multi-parameter assessment: albumin and urea secretion (markers of synthetic function), ALT/AST release (membrane integrity), CYP activity loss (metabolic function), and bile acid accumulation (cholestatic injury). Troglitazone, a drug withdrawn due to idiosyncratic hepatotoxicity, produces concentration-dependent toxicity in liver chips at clinically relevant concentrations (10–50 µM) that were missed in static hepatocyte assays.
7.2 Cardiotoxicity and Multi-Organ Systemic Toxicity
McAleer et al. demonstrated that a liver-heart multi-organ chip could reveal cardiotoxic effects of drug metabolites that were undetectable in static cardiac cultures. For example, the antihistamine terfenadine is metabolized by CYP3A4 in the liver compartment to its active form; when liver metabolism is functional, the metabolite produces QT prolongation and arrhythmia in the cardiac compartment. This integrated approach captures the critical role of metabolism in systemic toxicity.
8. Integrated ADMET: Multi-Organ-on-Chip Platforms
8.1 Systemic Pharmacokinetic Modeling
The ultimate goal of ADMET research is to predict human PK parameters from in vitro data. Multi-organ chips integrating gut, liver, kidney, and target tissues enable "clinical trial-on-chip" approaches where concentration-time profiles in each compartment mirror in vivo plasma and tissue concentrations. Novak et al. (2024) introduced a dynamic Microphysiological System Chip Platform (MSCP) integrating intestine, liver, heart, and lung compartments for comprehensive oral drug evaluation. Physiologically-based pharmacokinetic (PBPK) modeling uses OoC-derived parameters (Papp, CLint, Clr, Vd) to simulate human PK and predict drug exposure.
8.2 In Vitro-In Vivo Extrapolation (IVIVE)
IVIVE converts in vitro metabolic clearance to predicted in vivo hepatic clearance using scaling factors: CLh = CLint × scaling factor × fu, where fu is the unbound fraction in plasma. For OoC systems, the scaling factor accounts for the hepatocyte number per gram of liver and liver mass (typically 1.3 × 10⁸ hepatocytes/g liver; 1.5 kg liver mass). Well-stirred, parallel tube, and dispersion models are used to predict hepatic extraction ratio and oral bioavailability.
9. Validation and Regulatory Considerations
9.1 Benchmarking Against Clinical Data
OoC-based ADMET predictions must be validated against clinical PK data. Concordance correlation coefficients (CCC) >0.7 for predicted vs. observed clearance and bioavailability are considered acceptable for regulatory submissions. The FDA's Innovation and Quality (IQ) Consortium is actively developing validation frameworks for microphysiological systems in drug development.
9.2 Standardization and Quality Control
Critical quality attributes for OoC ADMET studies include: tissue viability (>80% for duration of study), barrier integrity (TEER within validated range), metabolic activity (CYP activity within 2-fold of fresh tissue), and reproducibility (CV < 20% for key parameters). Standard operating procedures (SOPs) for chip fabrication, cell sourcing, culture conditions, and analytical methods are essential for cross-laboratory reproducibility.
10. Conclusion
Organ-on-chip technology represents a transformative approach to ADMET assessment, enabling integrated, physiologically relevant studies of drug absorption, distribution, metabolism, excretion, and toxicity. By combining tissue-specific models within microfluidic platforms that recapitulate human physiology, researchers can generate more predictive ADMET data earlier in the drug development pipeline, reducing attrition and accelerating the delivery of safe, effective therapeutics.