Complex in vitro model: A transformative model in drug development and precision medicine

Abstract

In vitro and in vivo models play integral roles in preclinical drug research, evaluation, and precision medicine. In vitro models primarily involve research platforms based on cultured cells, typically in the form of two‐dimensional (2D) cell models. However, notable disparities exist between 2D cultured cells and in vivo cells across various aspects, rendering the former inadequate for replicating the physiologically relevant functions of human or animal organs and tissues. Consequently, these models failed to accurately reflect real‐life scenarios post‐drug administration. Complex in vitro models (CIVMs) refer to in vitro models that integrate a multicellular environment and a three‐dimensional structure using bio‐polymer or tissue‐derived matrices. These models seek to reconstruct the organ‐ or tissue‐specific characteristics of the extracellular microenvironment. The utilization of CIVMs allows for enhanced physiological correlation of cultured cells, thereby better mimicking in vivo conditions without ethical concerns associated with animal experimentation. Consequently, CIVMs have gained prominence in disease research and drug development. This review aimed to comprehensively examine and analyze the various types, manufacturing techniques, and applications of CIVM in the domains of drug discovery, drug development, and precision medicine. The objective of this study was to provide a comprehensive understanding of the progress made in CIVMs and their potential future use in these fields.

INTRODUCTION

Drug development plays an important role in medical advancements and directly affects the quality and processes of disease prevention and treatment. The complete process of drug development commences with early drug discovery, progresses through preclinical studies, human trials, regulatory reviews, and commercial approval, and culminating in the availability of the drug on pharmacy shelves, where postmarketing measures are implemented to monitor potential adverse drug reactions. ¹ This rigorous and costly investment, averaging 10–15 years and billions of dollars, underpins the extensive process that a new drug undergoes prior to official approval for clinical use. ² Drug development can be divided into preclinical studies and clinical trials, according to the experimental stage, objects, and models. Preclinical studies use various models to identify, screen, and test potential drugs, with only those drugs exhibiting efficacy and safety advancing to subsequent clinical trials. ³

In vitro cell cultures are useful research tools for modeling human diseases and offer a reproducible and rapid method for evaluating drug effects and safety. Whereas conventional 2D‐cultured monolayer cells in vitro have been widely used in recent decades, ⁴ superior 3D‐cultured cell models have been developed to achieve greater function and state of cultured cells. However, both these simple in vitro models do not fully replicate the complex tissue environment or the physiological and pathological processes that occur in the human body, and thus cannot mimic the physiologically relevant functions of human or animal organs and tissues.

Complex in vitro models (CIVMs) have been defined as systems in a 3D multi‐cellular environment within a biopolymer or tissue‐derived matrix, which incorporates primary or stem cell‐derived cells, immune system components, and mechanical factors, such as stretch or perfusion, or at least two of these elements. ⁵ CIVM encompasses frontier 3D cell culture, including organoid technology, organ bud culture, spheroid culture, tissue slice culture, 3D bioprinting, and hydrogel‐based tissue engineering. Another representative technique of CIVMs is microfluidic technology, which enables the precise manipulation of fluid flow to replicate blood circulation and further simulate drug absorption, distribution, metabolism, and elimination by fabricating interconnecting microchambers and microchannels. Consequently, CIVMs such as patient‐derived organoids (PDOs), ⁶ and organ‐on‐chip technology ⁷ are now more frequently used in disease research and drug screening. Compared to normal in vitro 2D cultured cell models, these CIVMs emulate the microarchitecture and functional characteristics of native organs and fully reflect the complexity of the drug response, thereby offering more accurate results for drug efficacy. Meanwhile, the “FDA Modernization Act 2.0” passed by the U.S Senate in 2022 authorized the use of certain alternatives to animal testing, including cell‐based assays and computer models, to investigate drug safety and effectiveness, eliminating the need for animal studies as part of the process to obtain a license for a biological product that is biosimilar or interchangeable with another biological product (https://www.congress.gov/bill/117th‐congress/senate‐bill/5002?q=%7B%22search%22%3A%22FDA+Modernization+Act+2.0%22%7D&s=10&r=1). ⁸ This signals a major shift in that animal tests are no longer indispensable, and CIVM may be an alternative to drug safety regulation.

CIVM has emerged as a promising approach for modeling disease, development, and homeostasis of various human organs and it plays an increasingly important role in drug screening and medicinal development. Understanding CIVM may help in the development of new drugs and precision medicine. In this review study, CIVM was introduced based on the progress of organoid technology, tissue slice technology, and microfluidic organ‐on‐chip technology (Figure 1). The applications of CIVM based on these technologies are also summarized to provide a reference for the development of CIVM.

FIGURE 1.

Schematic diagram of common in vitro models and typical CIVM (by Figdraw UWTTI9f63b). Cells used to establish in vitro models are derived mainly from cell lines or primary cells isolated from tissues. Single‐cell suspension can fabricate various in vitro models such as organoids, spheres, and microfluidic organ chips with the use of different culture technologies. Tissue slices can be formed from total tissue by tissue slicing. CIVM, complex in vitro model.

3D CELL CULTURE SYSTEM REPRESENTED BY ORGANOIDS

An organoid is a “3D structure derived from either pluripotent stem cells (PSCs), neonatal tissue stem cells or adult stem cells (ASCs)/adult progenitors, in which cells spontaneously self‐organize into properly differentiated functional cell types and progenitors, and which resemble their in vivo counterpart and imitating some functions of the organ.” ⁹ Although various 3D culture technologies have been rapidly established and widely used after the successful isolation of Matrigel in the 1980s, ¹⁰ the first single stem cell‐derived organoid, which has special characteristics of self‐proliferation capability and differentiation potential to form organ‐like structures, was not established until Sato et al. reported that Lgr5+ intestine stem cells self‐organized into intestinal crypt‐villus structures without a mesenchymal niche in Matrigel in 2009. ¹¹ This technique marked significant revolutionary progress in enabling strict recapitulation of in vivo cell signatures and has since emerged as a powerful tool for maintaining epithelial cells in a near‐native state. ¹¹

According to the literature, organoids may be generated through induced differentiation of stem cells. With advancements in human stem cell culture, organoids from various cellular sources, and species have been generated successfully. ¹² The three fundamental elements of organoid formation are media composition, cells, and the matrix (Figure 2).

FIGURE 2.

Three key points of organoid formation (by Figdraw, ID: OIYAI1111a).

Media composition for cultivation of organoids

For the development and growth of organoids, media compositions that recapitulate the in vivo stem cell niche signaling pathways that sustain stem cell function, drive their expansion, and eventually their differentiation, are essential. ¹³ Most organoids are derived from stem cells, which require exposure to specific morphogens at defined timepoints to activate the desired developmental signaling pathways and trigger self‐organization. This determines the importance and diversity of organoid media composition.

Various types of organoids require specific optimal media compositions, and only when components are provided exogenously can organoids proceed with appropriate generation and maintenance. Mouse optic cup organoids predominantly rely on endogenous signals, eliminating the need for specific exogenous signals and are cultured in a serum‐free medium with minimal growth factors. The formation of its uniform neuroepithelium, followed by self‐patterning mechanisms that specify spatially separated domains of the neural retina, retinal pigmented epithelium, and morphogenesis, all proceed appropriately and follow its default developmental trajectory. ¹⁴ However, the initial cellular system of most organoids lacks essential components to undergo the desired self‐organization process. The media composition of the majority of organoids requires supplementation with specific exogenous signals to ensure their intended and correct developmental trajectory. Takasato et al. ¹⁵ reported a protocol for the generation of kidney organoids from human embryonic stem cells (hESCs). Supplementation with specific growth factors, such as BMP4, activin A, FGF9, BMP7, and RA, was required to stimulate hESC differentiation to reciprocally induce kidney progenitor populations, which would then self‐organize into kidney organoids without further factors. Some organoids, such as gastric organoids ¹⁶ and human fetal‐like forebrain organoids ¹⁷ require specific exogenous stimulation throughout the derivation process. Supplementation of media composition may vary depending on the specific type of organoid. The contents of additional specific exogenous signals, such as growth factors, signaling agonists, and inhibitors also vary in organoid cultures. For example, Wnt/β‐catenin signaling, TGF‐β signaling, and EGFR2 signaling‐associated components, such as Wnt‐3A, BMP‐4, Activin A, EGF, FGF‐10, FGF‐7, HGF, and SB 431542, are included in the list of media supplements. ¹⁸ , ¹⁹

Cell resources for organoid generation

Two types of stem cells are used in organoid culture. One is embryonic pluripotent stem cells (ESCs) or ESC states, such as induced pluripotent stem cells, which are responsible for embryonic organ development. These cells have been successfully used to differentiate into and form a broad range of tissue‐specific organoids. ²⁰ , ²¹ The other type comprises ASCs or organ‐specific resident stem cells, which are important for maintaining mature organ homeostasis and facilitating regeneration. ²² These stem cells have already demonstrated their ability to grow into organoids in vitro after obtaining the proper extracellular matrix (ECM) and molecular clues. For example, mouse intestinal organoids are initially generated from Lgr5 + ASCs in the absence of a mesenchymal niche. ¹¹ In summary, most organoids from the surface ectoderm lineage or endodermal lineage, represented by glandular tissues, are derived from ASCs, dissociated adult tissues, or PSCs. Neuroectodermal, cerebral, and mesodermal kidney organoids are exclusively derived from PSCs. ²³

Generally, organoids can be obtained through the derivation of a single cell type or co‐culture of separate pre‐established cell types. However, the starting conditions and research potentials may differ. Organoids from single‐cell types, such as the reported optic cup or small intestine organoids, typically undergo an initial cell expansion step before self‐organization, ²⁴ whereas the co‐culture methods of organoids initiate from pre‐differentiation and an appropriate proportion of each cell type. ²⁵ Mesenchymal stem cells (MSCs) have previously been shown to contribute to the contraction force that drives self‐organization and are indispensable for the co‐culture approach of organoid formation. ²⁶ The scope of applicability of an organoid as a biological model system varies due to the differences in the formation process. Organoids derived from a single cell type that undergoes cell differentiation and the simultaneous generation of different cell types may offer additional insights for organogenesis. Organoids from co‐cultured pre‐differentiated cells establish distinct cell identities. Their self‐organization primarily involves cell sorting and subsequent architectural rearrangements, enabling them to comprehensively capture the transient developmental interactions between different progenitors during organoid formation. ²⁶

Matrix for organoid generation

The last element is the matrix, which supports cell growth and adherence. Matrigel, a natural solid ECM purified from Engelbreth‐Holm‐Swarm mouse sarcoma, ²⁷ is widely used to promote intestinal, cerebral, gastric, and mammary gland organoids. ²⁰ , ²⁸ Matrigel and other similar animal‐derived hydrogels, such as collagen type I matrices, mimic basement membranes and support cell adherence and growth. These natural matrices contain a complex mixture of ECM components and growth factors, facilitating efficient cell growth and differentiation. However, these matrices exhibit batch‐to‐batch variability, making it almost impossible to analyze and control their exact compositions and contents. This increases the difficulty in controlling the culture environment and reproducing experimental results. Chemically defined hydrogels have a definite composition and therefore represent promising candidates. Synthetic hydrogels, such as alginate hydrogels, ²⁹ hyaluronan‐gelatin hydrogel, ³⁰ methacrylate gelatin, ³¹ polyethylene glycol, and polyacrylamide ³² may support the culture of intestinal, pancreatic, neural, colorectal cancer, prostate cancer, glioblastoma, and hepatic organoids and retain the expression profile of key markers. ³³ , ³⁴ Further, these synthetic hydrogels allow the biochemistry and mechanics of the cultural environment to be controlled. However, they lack bioactivity and need to be customized to meet the specific requirements of different organoids.

With considerable advances in organoid culture technology, Matrigel is no longer an essential element. Co‐culture‐based organ buds are considered non‐matrix (Matrigel) organoids because they are also tiny self‐organized structures. These organ buds are assembled from various PSCs through intercellular and ECM interactions. The first reported co‐culture‐based organ buds were liver buds, which were created by mixing tissue‐specific progenitor cells derived from PSCs, endothelial cells (ECs), and MSCs. ²⁵ Through extensive research, organ buds for different organs, such as the kidneys, pancreas, intestines, heart, lungs, and brain, have been developed. ³⁵

Another strategy for growing organoids involves suspension cultures. This method has already been adopted for deriving optic cups, cerebral, cerebellar, kidneys, acinar/ductal, and hippocampal organoids. ³⁶ , ³⁷ The suspension method involves growing the organoids inside thermoformed microwell arrays and promoting their controlled growth under matrix‐reduced conditions, with Matrigel being used only as a medium supplement and not a scaffold, or even under no‐matrix conditions. Kumar et al. generated human PSCs‐derived kidney micro‐organoids using a modified suspension culture method. Briefly, cell suspensions in low‐adhesion culture plates were swirled at low‐speed (60 rpm) to form cell aggregates in the presence of differentiation media containing 0.1% polyvinyl alcohol and methylcellulose. However, the formation of variable patterns of dysplasia present in the prolonged suspension culture resulted in the loss of a functional proximal tubule, reduced expression of many kidney marker genes, and fibrotic lesions followed by apoptosis of the epithelium and deposition of the ECM (α‐SMA). ³⁷

Human colorectal epithelial organoids ³⁸ have also been generated using low‐viscosity matrix suspension culture. Organoids and toroids grown in a culture medium containing 5% Matrigel and 10 μM Y‐27632 appear to reduce the cost of organoids and fitting high‐throughput drug screening due to their ability for rapid expansion. However, the ECM is a fundamental, core component of all tissues and organs. The stiffness and biochemical substances of the matrix are important components of the microenvironment, and the matrix is involved in various biological behaviors. ³⁹ Cancer cells, the heterogeneous collection of infiltrating and resident host cells, secreted factors, and ECM together make up the tumor and tumor microenvironment. ⁴⁰ Increased matrix stiffness reportedly has profound effects on tumor growth and metastasis. The absence of a matrix when establishing organoids in vitro may compromise mimicry of the tumor microenvironment in vivo. ⁴¹ Thus, careful selection of matrix‐reduction or non‐matrix methods for organoids is vital to maintain the accuracy of drug screening.

TISSUE SLICING TECHNOLOGY AND TISSUE SLICES

The tissue slice culture prototype was first proposed by Dr. Harford's team and involved cutting the tissue into small pieces. ⁴² These tissue slices can be separated directly from normal organs and tumor tissues from humans and animals, providing the natural advantage of maintaining complex structural features and microenvironments in vitro. ⁴³ However, tissue slices cannot be maintained under long‐term culture or expansion in vitro, with the culture period typically lasting 1 to 2 weeks only. ⁴⁴ These characteristics make tissue slices suitable for the short‐term study of disease mechanisms and prediction of single‐use curative effects in vitro, especially for immune‐related drugs or therapies.

Fabrication methods of tissue slices

Tissue slices can be fabricated using two methods. The first method is manual cutting, which typically produces millimeter‐thick samples with irregular shapes. It is difficult to ensure the survival of cells at the center of the sample and the process lacks repeatability. ⁴⁵ The other method is vibratome technology, which can generate slices from tens to hundreds of microns. Tissue slices generated by the vibratome are precisely cut to obtain a relatively regular shape, making them superior for culture and testing. ⁴⁶ Therefore, the second method is most commonly used to generate tissue slices.

Equipment, such as a Krumdieck microtome, vibratome, and compresstome can used for tissue slicing. These equipments rely on the vibration of the blade during slicing to reduce pressure and stress on the fresh tissue sample. ⁴⁷ The Krumdieck microtome, vibratome, and compresstome differ in the thickness of the tissue that can be cut and the way the sample is sliced. The Krumdieck microtome generates slices ranging from ~100 to 500 μm, with a slicing rate of 3–4 s per slice. Before slicing, the tissue is cut into a cylindrical shape and placed into a sample hole, which is then fixed perpendicularly to the microtome blade. To regulate slice thickness, the distance between the blade and the screw‐controlled limit plate is adjusted during the slicing process. ⁴⁸ The compresstome shares a similar slice generation process to the Krumdieck microtome but the slice thickness ranges from ~30 to 1000 μm. The VT1200 S vibratome (the most commonly used vibratome) produces slices of a similar thickness to that of the compresstome. The tissue is immobilized on the sample table of the Leica VT1200 S vibratome before slicing, and the slice thickness is controlled by lifting the sample table.

Culture of tissue slices

Culture slices is followed by generation. The vitality and specific functions are the basis for further biological experiments. Slice thickness, culture medium, and specific culture methods are important considerations.

Research has determined that tissue slice thickness of 100–300 μm is optimal due to the limitation of oxygen and nutrient penetration, which typically spans 10–20 cell layers (~150 μm). Moreover, slices thinner than 100 μm may result in a high percentage of injured cells caused by slicing. ⁴⁵ Slices of 200–300 μm thickness are the most commonly reported in publications to date. In addition, the media used for these tissue slices are easier to use than that for organoid culture, considering that they do not require differentiation or self‐organization processes. The culture medium is similar to that of homologous cell culture. However, the culture of immune cells requires the appropriate addition of IL‐2, especially when testing immunomodulatory drugs. ⁴⁹ For specific culture methods, tissue slices are usually placed on a supporting tool, such as a Millipore filter insert, to navigate the air‐liquid interface culture. ⁵⁰ However, these methods still result in intra‐slice gradients. Microfluidic techniques have been used to overcome this limitation. The perfusion air culture system provides a continuous and controlled oxygen medium and drug supply. ⁵¹

MICROFLUIDIC TECHNOLOGY AND MICROFLUIDIC CHIPS FOR CIVM

The microfluidic technique involves the precise manipulation of fluids using microscale device technology first developed by the semiconductor industry and later expanded upon using microelectromechanical systems field. ⁵² These techniques are based on the combined principles of physics, chemistry, biology, fluid dynamics, microelectronics, and material science, and are characterized by engineered‐manipulated fluids at submillimeter‐scales. ⁵³ Precisely‐controlled fluids and particles enable the manipulation and analysis of cultured cells. Researchers recognize the significance and advantages of medicine and engineering across disciplines and concur that the development of comprehensive microfluidics could potentially solve issues in biology and clinical research. However, the integration of novel microfluidic techniques in mainstream biological research has not matched the initial enthusiasm in the field. Reports indicate that microfluidic technology has been primarily used for diagnostic applications and the manipulation of blood samples in mainstream biomedical research over the past decade, despite its potential in areas such as nanoparticle preparation, drug encapsulation, delivery, targeting, cell analysis, diagnosis, and cell culture. ⁵³ Fortunately, the development of cell culture technology has increased the demand for drug development and precision medicine. The adoption of novel microfluidic techniques in constructing in vitro models has progressed rapidly and has become the primary application of microfluidic technologies because the microfluidic system integrates the functions of driving, manipulating, monitoring, reacting, detecting, and analyzing. This allows for the simultaneous optimization of culture conditions, treatment, and detection. ⁵⁴ Microfluidic organ chips can be categorized into single‐organ chips and human‐on‐chips based on their target function, degree of complexity, and the types of organs in the system (Figure 3).

FIGURE 3.

Typical microfluidic organ chips (by Photoshop). (a) Diagram of a sandwiched human liver‐chip fabricated by primary human hepatocytes within an extracellular matrix on a porous membrane. ⁶⁹ The whole channel was divided into an upper parenchymal channel and a lower vascular channel by the sandwiched structure. (b) Diagram of a physiologically inspired two‐organ chip. ¹¹⁰ Liver spheroids and bronchial tissue were seeded in the right and middle chambers separately to mimic the MucilAir culture and liver model. The medium in the left chamber progressed through the MucilAir culture chamber and then through the liver chamber. (c) Diagram of a physiologically inspired four‐organs‐chip. ⁷⁰ This chip consists of a surrogate blood circuit (red) and an excretory circuit (orange). The surrogate blood circuit includes three chambers simulating the liver, brain, and the intestines, whereas the excretory circuit includes two chambers simulating the glomerulus and renal tubule.

Microfluidic organ‐on‐chips (single‐organ chips)

The use of microfluidic devices has considerably advanced the field of CIVMs because of their ability to mimic the fluid environment of native living cells and allow for the co‐culture of complex cellular components. ⁵⁵ Microfluidic organ‐on‐a‐chip or human‐on‐a‐chip is the most typical application and can potentially replace animal testing. ⁵⁶ The concept of the organ‐on‐a‐chip system originated from attempts to control and optimize cell culture in vitro by applying various microfluidic systems. The aim is to mimic the key organotypic cellular architecture, functionality, and environment on a smaller scale for the purposes of disease modeling and drug screening. ⁵⁷

The field of microfluidic organ‐on‐chip has rapidly progressed since the lung‐on‐a‐chip model was built on Huh's early work by the Takayama Group in 2010. ¹ , ⁷ This lung‐on‐a‐chip model had two microfluidic channels separated by a porous membrane on which lung alveolar and capillary cells could be co‐cultured. This model provided researchers with the opportunity to decipher the breathing mechanisms occurring in the alveoli, the capillary interface of the human lungs, environmental effects on lung cells, and the pathological mechanisms of various pulmonary or other respiratory diseases in vitro. ⁵⁸ To date, organ‐on‐chips for investigating disease progression and analyzing adverse drug reactions include liver chips, lung chips, heart chips, kidney chips, pancreas chips, gut chips, bone and bone marrow chips, brain chips, reproductive organ chips, and muscle chips have been successfully developed.

Biological barriers are important for the maintenance of organ homeostasis and are essential for drug testing and disease modeling. ⁵⁹ In vitro models of functional biological barriers, such as blood–brain barrier (BBB), ⁶⁰ have previously been created through the development of 3D cell culture techniques and microfluidic organ‐on‐chip technology. Yu et al. ⁶¹ developed 3D microfluidic BBB chips by co‐culturing rat primary brain microvascular ECs, pericytes, and astrocytes from neonates in a collagen matrix. Similarly, Kim et al. ⁶² constructed a simplified 3D co‐culture‐based BBB model within 30 min using immortalized human brain ECs and immortalized human astrocytes mixed with Matrigel. This simplified 3D co‐culture‐based BBB model, comprised solely of immortalized brain ECs, blocked the penetration of dextran molecules with various molecular weights, remained durable and impermeable even under BBB‐degrading conditions, and rapidly formed tight junctions.

Microfluidic human‐on‐chips

Currently, the more advanced and complex “body‐on‐chip” or “human‐on‐a‐chip,” mirrors the physiology of the entire human body for drug pharmacokinetic and pharmacodynamic analyses. This technology uses interconnected in vitro microfluidic devices to model human tissues and was established using multiple organs‐on‐chips. ⁶³

Building on the concept of single‐organ chips, more complex and advanced multi‐organ chips integrating multiple organ units have been constructed. Multi‐organ chips can mimic individual organ functions and have further advantages in integrating the functions of the constituted single‐organ part, such as drug absorbance in the gut compartment, drug metabolism in the liver compartment, and drug elimination in the kidney compartment. It mimics multiple systems that interact in vivo thereby enabling comprehensive studies. For example, Pires de Mello et al. ⁶⁴ developed a three‐organ heart–liver–skin system. A skin surrogate (Strat‐M membrane) was used to mimic the absorption processes of the topically administered drugs to be tested and assess their toxicity. The results indicated that the heart–liver–skin three‐organ system can be used to assess potential drug toxicity from dermal absorption, as well as evaluate transport dynamics through the skin. Shi ⁶⁵ developed a BBB‐glioma microfluidic chip (BBB‐U251 chip) composed of a BBB unit and glioma cells. The BBB unit was formed by the co‐culture of primary human brain microvascular ECs, pericytes, and astrocytes. The BBB‐U251 chip displayed selective permeability to FITC‐dextran with various molecular weights and three model drugs with different permeabilities. This glioma model could replicate the barrier function of the human BBB as well as the glioma microenvironment.

Human‐on‐chip confers the advantage of investigating inter‐organ communication in response to drug challenges at the human level, which provides drug development activities with the potential to lower the cost of preclinical studies and increase the rate of drug approval by introducing human phenotypic models early in the drug discovery process. ⁶⁴ It also provides a novel approach to rare disease research and orphan drug development. ⁶⁶ For instance, researchers at Cornell University created a 13‐organ recirculating system. This microfluidic cell culture device had pumpless 14 chambers (13 organs) that allowed for separation between the barrier and non‐barrier cell culture types, and supported all cell types maintained for up to 7 days. ⁶⁷

Key parameters for organ‐chips

One of the most important purposes of microfluidic organ chips is to provide cultured cells and tissues with a media flow similar to the circulatory system and fluid extracellular environment in vivo. Culture chambers and liquid control systems are critical constituents for achieving this. The design and regulation between the volume and runner of the chamber and the input and output of the medium lead to variations in the medium flow rate which may influence the culture of cells or tissues. ⁶⁸ Here, we summarize some culture parameters.

The design of chips varies greatly to fit different applications but have similar design logic; culture chambers and fluid chambers are essential structures and the key point of distinction. The two most common designs are strip‐type and round‐hole‐type culture chambers. For instance, the classical chip of Emulate Inc. has two strip‐type chambers (top and bottom channels) separated by a porous poly dimethylsiloxane (PDMS) membrane (Figure 3a). The cells are seeded on both sides of the PDMS membrane. ⁶⁹ The classical chip of TissUse GmbH has a round hole linked by different microfluidic channels (Figure 3b). Cells, organoids, and other samples are cultured in a round hole. ⁷⁰ As chips have similar culture chambers, they can achieve different functions and can be differentiated based upon the fluid flow direction and its cause. A classical type of flow is generated by pressure from a pump, enabling unidirectional fluid flow from the input to the output side of the culture chamber. Another type of flow is generated by sloshing the liquid. A chip plate containing an appropriate amount of medium is placed on a shaking device that moves similar to a seesaw. The fluid flows from a higher position to a lower position, generating a unidirectional flow, and then reversing as the shaking device changes position.

Various cells and tissues experience different fluid flow stresses in vivo; therefore, they have different optimal shear stresses and tolerance ranges. ⁷¹ Consequently, the flow rate is an important consideration because it can influence the fluid flow stress experienced by the cultured samples.

Materials used in microfluidic organ chips

The selection of materials for microfluidic chips plays a vital role in facilitating the utilization of microfluidic models in drug development, as it directly impacts the compatibility and interaction of drugs within these models. The decision regarding materials in microfluidics is of utmost importance, with biocompatibility being the primary consideration due to its significance in supporting the cultivation of biological cultures within microfluidic organ chips. Subsequently, the interaction with drugs or detection reagents must be taken into account as a secondary consideration. Materials with high binding affinity or drug adsorption properties with drugs and detection reagents can interfering with the results of drug screening or basic medicine research. Finally, the optical transparency, fabrication difficulty, mechanical properties, and cost also need to be considered. ⁷²

The PDMS is extensively utilized in microfluidic devices due to its transparency, ease of fabrication, and biocompatibility. Its diminished affinity for hydrophobic drugs renders it suitable for drug screening and delivery investigations. Nonetheless, PDMS has the propensity to adsorb hydrophobic drugs, leading to their gradual depletion from the solution. ⁷³ Glass, on the other hand, is a frequently used material that exhibits remarkable chemical resistance and optical transparency. Its limited drug adsorption properties render it appropriate for drug investigations that necessitate precise maintenance of drug concentrations. ⁷⁴ Polymethylmethacrylate, a cost‐effectiveness transparent material possessing favorable optical attributes and low drug adsorption properties, is widespread used in microfluidic devices and drug compatibility studies. Besides, materials like natural biomaterials (represented by gelatin, alginate, and collagen), PLGA, et al. are used to fabricate microfluidic chips. ⁷²

APPLICATION OF CIVMS IN DRUG DEVELOPMENT AND PRECISION MEDICINE

Positioning of CIVMs in drug discovery and drug development

Drug discovery focuses primarily on identifying and developing new therapeutic targets and potential drug candidates. This involves initial research and exploration to identify molecules or compounds that may potentially treat specific diseases, including activities such as target identification, lead‐compound identification, and early‐stage testing in laboratory settings. Drug discovery is primarily conducted by scientists in academic and industrial research laboratories. Once a promising drug candidate is identified during the drug discovery phase, it progresses to the drug development stage, which involves extensive testing and evaluation to determine its safety, efficacy, and optimal dosage. This stage includes preclinical studies, clinical trials (phases I, II, and III), regulatory approval processes, and postmarketing surveillance. Drug development focuses on refining and optimizing candidate drugs; understanding their pharmacokinetics, toxicity, and potential side effects; and gathering sufficient clinical evidence to support their approval and commercialization.

Animal and in vitro cell experiments have long been integral to life science research. Experiments in the drug discovery and development stages should be conducted on animals or in vitro 2D models that exhibit similar habits or characteristics that are relatively consistent with those of humans to study ontogeny, disease pathogenesis, and drug treatment effects. This is an essential step for new drugs before they are permitted to enter clinical trials. Only drugs with proven safety and efficacy, following a series of animal studies, are approved for clinical trials. ²

Millions of animal models have been established and are widely used in basic medical research. Small mammals, such as mice, rats, and zebrafish, as well as large mammals, such as swine and nonhuman primates, are popular and valuable in preclinical testing due to the similarity of their phenotypes, organ sizes, and physiologies to those of humans. Animal models are crucial for biomedical research and drug development. The translation of animal models into human subjects remains unpredictable owing to species variation and inherent heterogeneity, which may lead to unexpected results. Moreover, ethical concerns and the necessity of animal models are frequently mentioned by the scientific community. ⁷⁵ The tenet of the three R's (Replacement, Reduction, and Refinement) and animal welfare in animal research are emphasized throughout the animal experimental process. There is an apparent trend toward greater standardization and reduced reliance on animal models under the guidance of the three R's.

The 2D models have played a significant role in the past decades, not only in basic medical research, but also in drug discovery and development. A cancer cell line panel consisting of dozens of human cancer cell lines derived from different types of cancers was used for high‐throughput screening of drugs prior to selection for further preclinical assessment in xenograft models. Using a large panel of 77 colorectal cancer cell lines providing sufficient cell lines to represent each genetically defined subtype in primary cancers, a clear relationship between 5‐fluorouracil sensitivity and mismatch repair status in a subset was shown. ⁷⁶ Whereas 2D models are classic and contribute significantly to scientific research, they do not reappear in physiological tissue environments or in vivo physiology and pathology in the human body. Therefore, they cannot mimic the physiologically relevant functions of human and animal organs and tissues. In contrast, simple 3D‐cultured cell lines were developed as an optimized version of 2D culture. An investigation indicated that oxyresveratrol (OXY), an anticancer therapeutic agent, inhibits breast cancer cell proliferation. Although, OXY is an effective cytotoxic agent in 3D tumor models, its effect in 2D models is less pronounced. ⁷⁷

Another study tested the viability of 3D or 2D cultured malignant pleural mesothelioma (MPM) cell lines and analyzed the antitumor effects of cisplatin (CPDD) and pemetrexed (PEM). The results demonstrated that PEM alone and PEM combined with CPDD most effectively reduced MPM cell viability. The 3D culture models are important and superior in cancer studies and in vivo‐like drug testing. ⁷⁸ Considering that cells live in a complex microenvironment with various cell types, extracellular matrices, and physical stimuli, the aforementioned simple in vitro models are not ideal for accurately representing the drug response in the human body.

Research on humans cannot be conducted directly until sufficient evidence confirms the safety and efficacy of a drug or treatment. Therefore, advanced models are required to obtain evidence of safety and efficacy at a lower cost and in a shorter time. CIVMs are platforms to depict body reactions against drugs and may reduce investments and shorten the time for drug discovery and drug testing both in drug development and precision medicine owing to its superior in applying human physiology and neuropathology at an organ or whole‐body level, compared to normal 2D cell and animal models. In August 2022, the US Food and Drug Administration (FDA) granted approval for a clinical trial Investigational New Drug (IND) application for Sanofi's drug Sutimlimab. This approval marks a significant milestone as it is the first time that the FDA has approved a clinical trial IND application based solely on preclinical efficacy data obtained from human organ‐on‐a‐chip studies, in conjunction with existing safety data (https://www.nih.gov/news‐events/news‐releases/researchers‐create‐3‐d‐model‐rare‐neuromuscular‐disorders‐setting‐stage‐clinical‐trial). Additionally, in the same year, the United States enacted legislation that eliminated the stringent mandate for animal testing prior to FDA approval for new clinical trials.

Representative CIVMs, such as organoids, organ‐on‐chips, or human‐on‐chips, and tissue slices usually appear in publications on disease research as the principal focus of disease modeling. These technologies and products differ from each other, but have the potential for complementarity and collaboration for significant advancements. For example, organoids usually have self‐assembled structures and multiple‐cell components but lack mechanical factors, such as perfusion. ⁷⁹ Tissue slices have complex immune system components that are highly consistent with the microenvironment in vivo but are limited in nutrient transport and long‐term culture, and mimic interstitial flow. ⁸⁰ These deficiencies limit the ability of a CIVM to reach a stage that is more consistent with the original tissue. Fortunately, the application of microfluidic culture technology provides an opportunity to overcome these limitations, enhancing the culture efficiency and interstitial flow stimulation for organoids and tissue slices. ⁸¹

These advantages make CIVM not only highly valuable in studying disease mechanisms and target screening during the drug discovery stage, but also superior in evaluating safety and efficiency during the drug development stage. Meanwhile, drugs that can be evaluated are limitless in nature. In addition to innovative drugs, CIVMs have further opportunities for the efficiency and safety evaluation of generic drugs and excipients, as well as expanding the indication of drugs. The absorption, metabolism, and toxicity of ginsenoside compound K, a carbohydrate drug with numerous biological activities and physiological functions, have been successfully investigated using single‐organ and multi‐organ chips based on intestinal, vascular, liver, and kidney chips. ⁸² Liu et al. analyzed 870 liver chips to determine their ability to predict drug‐induced liver injury caused by 27 known hepatotoxic and non‐toxic drugs. With a sensitivity of 87% and a specificity of 100%, the liver chips outperformed conventional models. This suggests that widespread acceptance of CIVMs, such as organ chips, could decrease drug attrition and generate billions annually for the pharmaceutical industry through increased drug development and small‐molecule research and development productivity. ⁶⁹

CIVMs for the study of disease mechanisms and identification of drug targets in drug discovery

Organoids, as a component of CIVM, can serve as models of human diseases. A wide range of organoid‐based disease models that replicate genetic diseases, host–pathogen interactions, and cancer have been developed and have certain well‐known pathological features. For example, Welm et al. ⁸³ reported human xenograft‐derived organoid (PDXO) cultures from patients with endocrine‐resistant, treatment‐refractory, and metastatic breast cancers. These PDXO models have demonstrated high fidelity to their original tumors, and their drug responses are concordant with in vivo responses and could predict drug responses in patient‐derived xenografts. Precancerous pathologies encompassing endometrial hyperplasia and Lynch syndrome organoids, patient‐derived cholangiocarcinoma organoids, ovarian cancer organoids, and patient‐derived upper tract urothelial carcinoma organoids that capture disease diversity have also been established and will serve as powerful research models to aid in drug screening and discovery. ⁸⁴

Microfluidic organ chips containing cells or organs represent an alternative method for drug discovery. Sengupta et al. ⁸⁵ established a breathing lung‐on‐chip system composed of immortalized human alveolar epithelial cells that represented both AT1 and AT2 characteristics, thereby presenting a valuable in vitro tool for studying inhalation toxicity, testing the safety and efficacy of drug compounds, and characterizing xenobiotics. Moreover, it has potential applications in coronavirus disease 2019 research. ⁸⁶

CIVMs for the study of preclinical evaluation in drug development

Disease models, such as PDOs or tissue slices, can serve as excellent tools for evaluating drug efficiency. Hong et al. fabricated tumor tissue slices from patients with clear cell renal cell carcinoma (ccRCC) to evaluate the antitumor effectiveness of a V‐domain Ig suppressor of T‐cell activation (VISTA) inhibitor, which is a candidate immune checkpoint inhibitor (ICI). ⁸⁷ The results indicated that all tested samples responded to the anti‐VISTA monoclonal antibody and produced TNF‐α, with 20% of ccRCC samples showing a synergistic effect when treated with a combination of VISTA and PD‐1 inhibitors.

Microfluidic organ chips are advantageous for studying cell interactions and testing drug candidates for diseases. Organ‐on‐chips (especially multi‐organ chips or human‐on‐chip models) are superior for assessing drug toxicity in healthy human organs and the efficiency of disease by considering drug absorption and metabolic processes in the body. ⁸⁸ A collagen‐based 3D primary human hepatocyte (PHH) model was constructed using the biomimetic array chip. This model showed improved and stabilized liver functionality in terms of cell viability, albumin, and urea production compared with the 2D PHH model and showed a higher sensitivity for predicting the hepatotoxicity of clinical drugs, indicating its potential for risk assessment of drug‐induced hepatotoxicity. ⁸⁹ Heart‐on‐chip and cardiac organoids have been used to predict cardiotoxicity of drugs. ⁹⁰ An optimized microfluidic chip design consisting of interconnected compartments was presented. Such a simplified tandem system is a liver‐kidney‐on‐chip model that includes a liver compartment containing hepatic cells that grow abundantly in microfluidic conditions and stably express metabolism‐related biomarkers and a glioblastoma compartment. The biotransformation and toxicity of Aflatoxin B1 and benzo(a)phapyrene, as well as the interaction with other chemicals, were successfully investigated using this system, demonstrating that the toxicity and metabolic response to drugs can be evaluated in advanced interconnected multi‐organs chips. ⁹¹ Intestine‐liver‐on‐chip systems have also been developed and used to predict oral drug administration and first‐pass metabolism in vitro as an emulation of the first‐pass mechanism occurring in vivo. ⁹² In addition, a four‐organ chip integrated with sequentially connected intestinal, liver, skin, and kidney compartments with stable homeostasis across different organ compartments was developed to test the heart and liver toxicity of acute and chronic drug exposure. ⁹³

Biological barrier chips are special well‐developed organ chips. Researchers modeled neuroinflammatory conditions that compromised BBB functionality and observed the protection of the BBB after treatment with the glucocorticoid drug dexamethasone. ⁶¹ In addition, in vitro BBB‐glioma microfluidic chip model could reproduce the high level of barrier function of the in vivo human BBB and allow for the establishment of the glioma microenvironment. ⁶⁵ The drug efficacy of six potential anti‐glioma components from traditional Chinese medicine was evaluated by delivery into the blood channel of the chip. The data closely resembled in vivo data from the traditional Transwell model, indicating that the effect of the drugs on glioma (U251) cells in the chip was significantly lower owing to the presence of the BBB. ⁶⁵ Despite significant progress in constructing BBB models, there is still a long way to go to replicate the BBB in vivo, as it is a dynamic multicellular interface that regulates the transport of molecules between blood circulation and the brain parenchyma. ⁹⁴ The cell type and degree of tight junctions in ECs are the most important factors determining the success or failure of the BBB model. They affect the transport of related substances between the endothelium and epithelium in the model because different cells obtain different pumps, which may lead to variations in transport and allow diverse substances to pass through. Constituent cell types, including non‐fenestrated brain microvascular ECs, microglial cells, pericytes, astrocytes, and neurons, play indispensable roles in BBB integrity. ⁹⁵ Other components, such as tight junction proteins, adherens junctions, and junctional proteins, may influence barrier permeability. ⁹⁶ None of these compositions can be completely simulated using a simple layer of the epithelium.

Finally, the integration of technologies mentioned above promotes the progress of CIVMs. Compared with 2D cultured cells, organoid techniques may improve the performance of organ chips. A 3D proximal tubule model, composed of epithelial cells isolated from kidney organoids matured under static conditions, exhibited significant upregulation of OCT2 and OAT1/3 transporters compared to that of control chips based on immortalized proximal tubule epithelial cells and was used to mimic basolateral drug transport and uptake in drug screening and disease modeling. ⁹⁷

CIVM for precision medicine

With the development of medical science, the range of drugs and treatment options for patients and doctors has grown. However, due to patient‐specific reactions and drug side effects, particularly in the treatment of tumors and rare diseases, there is growing concerns regarding precision medicine that could match patients' individualized medication needs. ⁹⁸ Consequently, CIVM fulfilled the demand for precision medicine.

Welm et al. ⁹⁹ developed tumor organoids from a 43‐year‐old patient with triple‐negative breast cancer, screened a library of FDA‐approved and experimental drugs, and found that eribulin and talazoparib emerged as promising candidates, whereas several of the chosen clinical therapies did not appear to be effective. This result matched the clinical treatment result, showing that the patient experienced early metastatic recurrence in the liver after serial treatment. Liver metastases and ascites regressed completely after eribulin treatment. Wang et al., ¹⁰⁰ developed PDOs and found that they mirrored patient clinical responses to platinum chemotherapy and displayed drug response heterogeneity to targeted agents, including PARPis. Additionally, they found that the use of combination strategies targeting the resistance mechanisms of a patient who relapsed during olaparib maintenance therapy could reverse the effects. Besides PDOs, including cervical cancer, head and neck squamous cell carcinoma, prostate cancer, lung cancer, rectal cancer, and endometrial cancer were developed as platforms to determine the effects of chemotherapy, radiation therapy, and targeted therapy in patients.

In addition to traditional chemical and targeted drugs, tumor immunotherapy has emerged as a new hope for cancer treatment. Immunotherapy is used to control and eliminate tumors by restarting and maintaining the tumor‐immune cycle and restoring anti‐tumor immune response. ¹⁰¹ Thus, immunotherapy is expected to have a higher demand for individualized drug selection. Furthermore, normal tumor PDOs and organ chips cannot be used to test the efficacy of tumor immunotherapy. Therefore, CIVMs, such as tissue slices, immune co‐culture organoids, and organ chips, may be an alternative.

Dijkstra et al. expanded autologous peripheral blood mononuclear cells (PBMCs) and established autologous tumor organoids. PBMCs were then stimulated weekly with tumor organoids to induce tumor‐specific T‐cell responses. This co‐culture system enriched the tumor‐reactive CD8+ T cells from PBMCs. ¹⁰² Tumor reactivity of the expanded autologous CD8+ T cells was analyzed. Organoid systems can be used to evaluate the efficiency of immune‐related tumor destruction. ¹⁰² Zhou et al. established a 3D co‐culture system consisting of patient‐derived cholangiocarcinoma organoids and immune cells (such as PBMCs or purified T cells) and studied organoid cell death caused by PBMCs and purified T cells. Distinct responses of different PDOs to direct and indirect contact with immune cells were noted. ¹⁰³ These results indicate that CIVMs with immune cells is technologically feasible and has great applications in assessing specific reactions to immunotherapies in vitro.

Based on the development of microfluidic and immune co‐culture techniques, CIVM‐containing immune cells have shown great potential in predicting patient reactions to immunotherapies. ¹⁰⁴ The efficiency of ICIs such as PD‐1 and PDL‐1 antibodies was determined using such a model. ¹⁰⁵ Studies on high‐grade serous ovarian cancer (HGSC) have indicated that bispecific PD‐1/PDL‐1 antibody showed superior efficiency compared to monospecific anti‐PD‐1 or anti‐PD‐L1 antibodies in HGSC organoid/immune cell co‐culture models. ¹⁰⁶ Primary chordoma PDOs have been generated and used to test the dose‐dependent effects of nivolumab in predicting treatment responses. ¹⁰⁷ Besides, Ou et al. generated melanoma PDOs and determined that αPD‐1 can reinvigorate CD8+ T cells and then induce melanoma cell death. ¹⁰⁵ Junk et al. fabricated an ex vivo non‐small cell lung cancer tissue slice model that maintained the morphological characteristics of tumor specimen for at least 12 days and maintained T‐cell function for 10 days in vitro. ¹⁰⁸ Further, the tumor‐killing effect and T cell responses to nivolumab (a PD‐1 antibody) treatment in this model were evaluated, and the results were compared with paired clinical outcomes. The two groups of tissue samples were successfully correlated with their clinical outcomes. Adoptive cell transfer therapies, such as tumor infiltrating lymphocyte transfer, chimeric antigen receptor (CAR) T cell transfer, and CAR NK‐cells transfer were also determined using CIVM‐containing immune components.

DISCUSSION AND CONCLUSION

CIVM represents an alternative tissue model with biomimetic human pathophysiology, effectively bridging the gap between animal studies and clinical trials. It can help identify critical biological mechanisms as well as test drug efficiency and toxicity in target organs at the preclinical development stage, thus providing a reliable reference for clinical trials in the drug development pipeline. Moreover, CIVM‐like PDOs or patient‐derived organ‐on‐chips can contribute to precision medicine, by evaluating the personalized sensitivity of clinical treatment, exploring mechanisms of resistance, and identifying effective strategies to address human heterogeneity.

The platform integrates different technologies, such as in vitro cell culture, organoid, co‐culture, and microfluidic techniques, and serves as a product of multidisciplinary and interdisciplinary science. It enables the monitoring of drug metabolism pathways and toxicity effects on target cells/organoids, while enhancing the efficacy and reliability of the experimental outcomes. Additionally, the combination of drug sensitivity tests and gene sequencing/analysis can facilitate appropriate and personalized treatment strategies for patients. ¹⁰⁰

Despite the rapid and sustained progress in the design and construction of biomimetic CIVM, some limitations that hinder its widespread application remains. The first limitation is that CIVM, such as organoids, tissue slices, and organ‐on‐chips do not fully recapitulate the microenvironment and fall short in completely emulating the complex physiology of organs or the human body. As a refined experimental model, even the most complex multi‐organ chips are far from real human organs and systems in terms of structures and integrated functions. Interpretation of results obtained by CIVMs may inadvertently overlook safety signals that could be indicated in animal studies. This may be the key issue limiting the use of the CIVM for preclinical evaluation instead of animal experiments. Second, there are still some drugs or treatments (for example, tumor immunotherapy and anti‐angiogenic drugs) that are difficult to predict, although numerous organoids and organs‐on‐chips have been widely studied and established to recreate patient responses to drugs. Tumor immunology and angiogenesis are intricate, multifaceted events that are harmonized by several organs or systems in vivo and need precise regulations. ¹⁰⁹ Tissue slices and immune co‐culture might be optimal choices to reproduce the immune microenvironment in vitro. However, disparities in the detection index to evaluate the model's response to immunotherapeutic drugs remain challenging. The third limitation is the absence of standardized and quality control criteria for materials (such as cells, tissues, proteins, and microfluidics) used in CIVMs. Cell culture practices vary in different laboratories, and the vitality of primary tissues or cells varies in different patients. This causes instability in the materials used in CIVMs. Moreover, because the formation and application of CIVMs are still in the early research phase, determining the most appropriate parameter conditions remains challenging. In addition, these factors make it difficult to formulate universally recognized standards and quality control criteria, which are essential to ensure the reproducibility and robustness of CIVM for intended applications in drug discovery and development. Thus, CIVMs need to be designed and improved based on experimental data.

Fortunately, novel technologies, such as cell sheet engineering, 3D bio‐printing, and even four‐dimensional (4D) bio‐printing, are rapidly being developed as powerful tools. Overall, the advancement of CIVM and the establishment of an appropriate evaluation index for drug assessment will underscore the suitability of CIVMs for the evaluation of various drug characteristics and administration modes, thereby achieving a more comprehensive and extensive assessment of the drug or treatment in vitro.

AUTHOR CONTRIBUTIONS

L.W., D.H., and J.X. contributed equally to this work. L.W., resources, writing – review and editing, visualization. D.H. resources, writing – original draft, writing – review and editing. J.X. writing – revise manuscript writing. J.H. conceptualization, resources, writing – review and editing, supervision. Y.W. conceptualization, resources, writing – review and editing, visualization, supervision. All authors have read and agreed to the final manuscript.

FUNDING INFORMATION

This study was supported by The National Key Research and Development Program of China (2022YFC2407303); Major Science and Technology Projects of Zhejiang Province (2020C03058); Research Center for Lung Tumor Diagnosis and Treatment of Zhejiang Province (JBZX‐202007).

CONFLICT OF INTEREST STATEMENT

The authors declared no competing interests for this work. D.H. is a lab researcher of Hangzhou Chexmed Technology Co., LTD. Y.W. is chief executive officer of Hangzhou Chexmed Technology Co., LTD.

Wang L, Hu D, Xu J, Hu J, Wang Y. Complex in vitro model: A transformative model in drug development and precision medicine. Clin Transl Sci. 2024;17:e13695. doi: 10.1111/cts.13695