Table of Contents

Definition

Microphysiological systems (MPS) are bioengineered in vitro platforms that integrate microfluidics, human-derived cells, and engineered biomaterials to reproduce the dynamic structural, biochemical, and mechanical microenvironment of living tissues and organs at the microscale.¹ These platforms, frequently termed organs-on-chips, contain living cells cultured under dynamic fluid flow and are designed to recapitulate key features of human physiology and pathophysiology with high precision.² An MPS typically comprises biofabricated three-dimensional tissue constructs integrated into a microengineered device that mimics the tissue microenvironment and provides functional outputs, and the technology draws simultaneously upon cell biology, tissue engineering, and microengineering to recapitulate organ-level physiology.³

The biological basis of an MPS is the organ-level functional unit rather than the whole organism. Individual systems model structures such as the alveolar-capillary interface of the lung, the hepatic sinusoid, and other organotypic units by reconstructing their tissue-tissue interfaces, multicellular composition, and solute transport behavior.⁴ Structurally, MPS devices employ microfluidic chambers in which perfusion generates physiologically relevant shear stress and soluble-factor gradients, while porous membranes coated with extracellular matrix support the adhesion and polarization of contiguous tissue layers; in mechanically active organs, cyclic strain applied by the device can reproduce organ motions such as breathing.⁴ The field is conventionally subdivided into single-organ systems and multi-organ systems that couple two or more tissue equivalents within a shared perfusion circuit.²

Purpose

The primary value proposition of MPS technology is the generation of preclinical evidence that reflects human biology directly, rather than by cross-species extrapolation. Conventional preclinical packages built on two animal species demonstrably failed to flag drug-induced liver injury that later emerged in patients, which is the failure mode that motivated large-scale performance assessment of MPS in predictive toxicology.⁵ By using human primary cells, human iPSC-derived cells, and patient-derived tissues, MPS avoid species differences in metabolism, transport, and toxicity and can reveal human-relevant toxicological effects earlier in development.^{1 2} The corresponding improvement in predictive validity carries direct economic value: modeling of routine human liver-chip adoption estimated an increase in small-molecule research and development productivity of approximately three billion United States dollars annually.⁵

Beyond predictivity, MPS advance the ethical and regulatory objective of replacing, reducing, and refining animal use in biomedical research.¹ The United States FDA Modernization Act has accelerated the transition from animal-based testing toward human-relevant new approach methodologies in drug development, shifting attention from technological feasibility to regulatory implementation; among these methodologies, MPS are recognized as capable of generating evidence that more directly reflects human physiology.⁶ MPS therefore serve as a bridge between conventional in vitro culture and clinical observation, improving the translation of preclinical findings into human trials while reducing dependence on animal models.³

Methodology

MPS devices are most commonly fabricated by standard soft lithography and replica molding of polydimethylsiloxane (PDMS), an oxygen-permeable, transparent, and biocompatible elastomer, with microfluidic networks formed against master molds; thermoplastic alternatives such as cyclic olefin copolymer and polystyrene are increasingly used to mitigate material-related artifacts and improve manufacturability.^{7 8} The canonical architecture consists of parallel microchannels separated by a porous extracellular-matrix-coated membrane, with parenchymal cells seeded on one side and vascular or stromal cells on the other, reconstituting a perfused tissue-tissue interface under controlled mechanical strain.⁴

Cell sources span primary human cells, human-induced pluripotent stem cell (hiPSC) derivatives, and self-organizing organoids. For example, hiPSC-derived human liver organoids seeded into microfluidic chips matured under perfusion and maintained liver-specific functions, including strong albumin secretion and cytochrome P450 activity, over 28 days of culture.⁹ In multi-organ configurations, microchannels interconnecting organ chambers are perfused by integrated pneumatic micropumps, and autologous hiPSC-derived organ equivalents have been co-cultured in a common, growth-factor-deprived recirculating medium, maintaining phenotypes of intestine, liver, and neuronal models for 14 days of continuous coculture.⁷ Readouts are increasingly obtained through integrated biosensors, real-time impedance and optical monitoring, and conventional immunoassays and imaging, which convert tissue phenotype into quantitative, time-resolved signals.²

Performance

Empirical performance has been characterized most extensively for human liver-chips. In a blinded evaluation of 27 small molecules selected because they evaded conventional preclinical safety workflows, a commercial human liver-chip displayed a true positive rate of 87 percent and a specificity of 100 percent when two hepatocyte donors were used and data were corrected for drug-protein binding, with a Spearman correlation of 0.78 against an independent clinical severity scale; these results were additionally verified by two external toxicologists.⁵ Economic modeling on the same data indicated that incorporating the liver-chip into existing workflows could reduce the proportion of toxic drugs entering first-in-human clinical trials by 11.3 percent.⁵ Under chronic culture conditions, hiPSC-derived liver organoid chips sustained albumin secretion and CYP450 activity for 28 days, and autologous four-organ chips sustained multi-tissue homeostasis for 14 days, establishing benchmark windows for long-term and repeat-dose experimentation.^{9 7}

Current technical boundaries remain material and systemic. PDMS absorbs hydrophobic small molecules and releases them over time, distorting pharmacokinetic and pharmacodynamic measurements and producing inconsistent dosing, and its gas permeability causes medium evaporation and complicates the maintenance of defined oxygen gradients; cyclic olefin copolymer reduces but does not eliminate sorption artifacts.⁸ Reproducibility is further constrained by Matrigel batch variability, inconsistent protocols across laboratories, and limited inter-institutional comparability, and the absence of full immune, endocrine, and systemic complexity restricts modeling of whole-organism responses.¹⁰ Regulatory readiness varies substantially across applications, and adoption is therefore expected to proceed incrementally, with MPS initially supporting defined decisions as complementary evidence rather than replacing conventional studies outright.⁶

Applications

The principal application domains are drug development, disease modeling, and precision medicine. In drug development, MPS are applied to drug-induced liver injury prediction, absorption, distribution, metabolism, excretion, and toxicity profiling, and pharmacokinetic and pharmacodynamic studies, with liver platforms among the most routinely modeled and cost-effective systems.^{5 3} Disease modeling applications include musculoskeletal pathologies, inflammatory and infectious disease research, and aging, reflecting the capacity of MPS to reproduce organ-level pathophysiology that static culture cannot capture.^{3 2} In precision oncology, patient-derived organoid-on-chip platforms support high-throughput drug screening and patient-specific response prediction, with reported clinical validation correlations of approximately 0.4 to 0.6 between organoid drug responses and patient outcomes, indicating partial predictive capacity.¹⁰

Standardized contexts of use are emerging through industry consortia, whose qualification guidelines defined the compound set and decision criteria used in the liver-chip performance assessment.⁵ The United States Food and Drug Administration has published a strategy for reducing animal testing in preclinical safety studies that explicitly names organ-on-chip and other new approach methodologies, and its scientists participate in cross-company MPS evaluation collaborations with the pharmaceutical industry.¹¹ Deployment is presently at the stage of research-validated, early industrial adoption: MPS data are entering pharmaceutical decision-making as fit-for-purpose complementary evidence, most maturely for hepatotoxicity, while full replacement of animal studies awaits broader characterization of performance and uncertainty across contexts of use.^{6 5}

Variations

• Organ-on-a-Chip — A single-organ MPS that reconstructs one organ-level functional unit, such as the alveolar-capillary interface or the hepatic sinusoid, within a perfused microfluidic device; it is the foundational and most widely deployed MPS configuration.^{4 2}
• Multi-Organ-on-a-Chip — Also termed body-on-a-chip, this variation interconnects two or more miniaturized organ equivalents within a shared recirculating perfusion circuit, enabling systemic pharmacokinetic, multi-organ toxicity, and inter-organ communication studies.^{1 7}
• Organoid-on-a-Chip — A hybrid configuration in which self-organizing stem cell-derived organoids are integrated into microfluidic chips, combining the architectural fidelity of organoids with perfusion-controlled, dynamic microenvironments and supporting extended culture with stable organ-specific function.^{9 10}
• Tumor-on-a-Chip — A disease-directed MPS that models tumor microenvironments, cancer immunology, and metastasis using patient-derived tumor organoids and perfused vascular compartments, serving immuno-oncology and personalized drug response prediction.¹⁰