Novel emerging cell and organoid systems for the study of drug metabolism and toxicity in humans

Abstract

The drug discovery and development process faces significant challenges, including high attrition rates and substantial financial investment, in part due to the limitations of traditional 2-dimensional (2D) cell culture systems and animal models to predict human drug metabolism, efficacy, and toxicity. This review highlights the emergence of novel in vitro human cell culture and organoid systems, such as 3-dimensional (3D) cultures, self-assembling organoids, induced pluripotent stem cell–derived models, and microphysiological system or organ-on-a-chip systems, as transformative solutions to the issues raised when extrapolating from 2D cell culture. These advanced platforms offer enhanced physiological relevance by better recapitulating complex in vivo microenvironments, thus improving the predictability and accuracy of preclinical drug assessment. In this study, we systematically cover the utility of these advanced systems in studying drug metabolism and toxicology across key organs like the liver, intestine, and kidney, emphasizing their advantages over conventional models in terms of cellular diversity, architectural complexity, and long-term functional maintenance. We also discuss the potential of integrating these novel systems into the drug development pipeline, particularly their compatibility with high-throughput screening and their alignment with the 3Rs principle (replacement, reduction, and refinement) for ethical research. Despite their immense promise, challenges remain; including the lack of standardized protocols, the complexity of data analysis, and the need for further advancements in vascularization, innervation, and immune component integration. We conclude by exploring future directions, including the crucial role of artificial intelligence and machine learning in analyzing complex datasets and the potential for personalized medicine through patient-derived organoids. Overcoming these challenges will be vital for these innovative platforms to revolutionize pharmaceutical development, leading to safer, more effective, and more efficiently produced pharmaceuticals.

Significance Statement

This article reviews the design, construction, and implementation of novel in vitro cell culture and organoid systems for preclinical drug metabolism and pharmacokinetics and toxicology studies. As such, it serves as a resource for interested parties who would like to learn about, and implement, these cutting-edge technologies into their drug discovery and development workflow.

1. Introduction

The process of drug discovery and development is characterized by significant financial investment, extensive timelines, and a high rate of attrition. Approximately 90% of drug candidates fail to advance from phase I clinical trials to market approval.¹ This challenge is particularly acute for antineoplastic agents, which exhibited an even lower success rate of 3.4% between 2000 and 2015.² A substantial proportion of these failures, specifically 52% due to lack of efficacy and 24% due to safety concerns during the 2013–2015 period, occur in the costly late-stage clinical phases (II and III), imposing a considerable financial burden on pharmaceutical companies.³ These costly failures have led to the paradigm of fail early and fail fast, demanding the need for early decision making in the drug discovery process. This paradigm has been heavily reliant on in vitro model systems, particularly those involving cell culture.^4,5 Since the early days of in vitro metabolic and toxicity studies, it has been appreciated that model 2-dimensional (2D) cell culture systems cannot adequately recapitulate all aspects of metabolism and toxicity that are observed in vivo. Issues of cellular transport, protein binding, and cell-to-cell communication have all plagued in vitro-to-in vivo extrapolation (IVIVE) for decades and have made some in vitro cell culture work of dubious value.⁶ While offering simplicity, speed, cost-effectiveness, versatility, and reproducibility, conventional 2D monolayer cell cultures grown on planar, rigid plastic surfaces lack the intricate physiological microenvironment and 3-dimensional (3D) cellular architecture characteristic of in vivo tissues.⁷ It is often the case that these models fail to adequately replicate crucial factors such as the extracellular matrix (ECM), complex cell-cell interactions, and physiological oxygen and nutrient gradients that are typically found in vivo.^7,8 The unnatural environment of the plastic culture dish can alter cell morphology, polarity, growth kinetics, and gene expression, leading to results that do not accurately reflect the disposition of drugs in the human body.⁹ In a 2D culture, cells have uniform and unrestricted access to nutrients, oxygen, metabolites, and signaling molecules, which is not the case in the heterogeneous microenvironment of tissues in the whole body. These important factors can significantly influence drug penetration and bioavailability. Additionally, the lack of proper cell-to-cell communication and interactions with the ECM in 2D cultures can also affect drug transport, metabolism, and overall cellular behavior.^10,11 Each of these issues by themselves, or more typically in combination, can lead to altered drug response and poor prediction of metabolism and toxicity. These deficiencies result in limited physiological relevance of the cell culture model and often lead to inaccurate IVIVEs, contributing to both false-positive and false-negative results during the preclinical drug screening process.^6,12 This, in turn, can result in high failure rates in later stages of drug development when these candidates are tested in animal models or human clinical trials.^1,13

Despite these drawbacks, cell culture remains a critical methodology for obtaining a primary assessment of the metabolic and toxicity pathways for most drugs and xenobiotics. This has recently led to the development of a number of new cell-based technologies, such as sophisticated 3D cultures, self-assembling organoids, and microphysiological system (MPS), or organ-on-a-chip (OoC), systems which have revolutionized the predictive power of in vitro preclinical studies.14, 15, 16 The fundamental objective of these novel systems is to more faithfully recapitulate the complex physiological functions, intricate tissue architectures, and dynamic cellular microenvironments that are present in vivo.¹⁷

By providing a more physiologically relevant context, these advanced models offer enhanced predictability and accuracy in evaluating drug efficacy and toxicity, thereby significantly bridging the translational gap between preclinical findings and clinical outcomes in humans.¹⁸ The adoption of these novel systems also aligns with the ethical imperative of the 3Rs principle (replacement, reduction, and refinement), actively minimizing the reliance on animal testing in drug development.^15,19 Moreover, new guidance from the Food and Drug Administration (FDA) has emphasized the need for nonanimal model (NAM) systems to be used in the preclinical stages of drug testing. Indeed, the FDA’s goal is to eliminate most animal testing by 2030.²⁰ However, the impetus behind this paradigm shift in model cell culture systems is not solely driven by the pursuit of enhanced biological accuracy or regulatory compliance; it is equally propelled by a pragmatic need for more efficient and ethically sound research methodologies. These advanced in vitro models offer compelling advantages such as greater cost-effectiveness, faster experimental turnaround times, inherent compatibility with high-throughput screening (HTS) methodologies, and a significant reduction in ethical concerns when compared with traditional animal models.²¹ The consistent emphasis on HTS capabilities across various advanced models indicates that these systems are not merely tools for in-depth mechanistic studies but are also highly effective for broad, early-stage compound filtering. This capability is crucial for accelerating the initial phases of drug discovery by efficiently narrowing down promising lead candidates. Consequently, the adoption of novel in vitro cell culture models represents a strategic evolution aimed at holistically improving the entire drug discovery pipeline, rendering it more predictive, economically viable, and ethically responsible.

In this review, we will cover typical 2D cell culture systems and their current use in ADME/tox (ADMET) studies, describe the newer types of cell culture systems—including induced pluripotent stem cell (iPSC) and organoid systems, elucidate the advantages of these systems over traditional 2D cell culture models, detail the potential utility of MPS systems in preclinical metabolism and toxicology studies, discuss integrating these novel systems and the data generated from them into the drug development pipeline, and finish by describing future directions of the field in 3D bioprinting and roles for artificial intelligence (AI) and machine learning (ML) in the preclinical ADMET paradigm. Not only do these new technologies promise the possibility of vastly improved IVIVE, but they also will streamline the drug discovery and development pipeline by reducing costs, minimizing the use of animal models, and providing candidate assessment in a more timely fashion.

2. Brief historical perspective: traditional cell-based systems for drug metabolism and toxicity testing

A few classic cell culture systems have been historically used to study drug metabolism and toxicity in the liver, intestine, and kidney (Fig. 1). In this section, we will offer brief descriptions of the most commonly used primary cells and cell lines and their employment in preclinical ADMET studies.

Fig. 1.

Traditional methodologies and cell types involved in drug metabolism and toxicity studies.

2.1. Human hepatic cell models

Primary human hepatocytes (PHHs) have long served as the gold standard in vitro cell culture model system for drug metabolism and hepatotoxicity studies.²² However, their limited availability, cost, short lifespan, and de-differentiation represent major barriers for accurate metabolism and toxicity testing. Traditionally, PHHs have been cultured in 2D monolayers, with or without an ECM/Matrigel overlay, or suspension incubations for short-term culture in order to provide initial estimates of drug metabolism and potential biotransformation-mediated hepatotoxicity. However, in 2D culture, PHHs have been observed to de-differentiate and lose significant metabolic and transport activity over time.^22,23 This phenomenon has become more significant in recent years due to the fact that a large number of new chemical entities have soft metabolic sites that have been blocked by nonmetabolizable functional groups, such as fluorine or cyclopropyl moieties, which can have the effect of extending the drug half-life several fold, necessitating the need for extended cell culture assays.24, 25, 26 Therefore, more recent optimizations of culture conditions have aimed to enhance the long-term use of hepatocytes, employing new media formulations and inhibition of mechanical tension that can support hepatocyte culture for up to 1–2 months.^23,27,28

In response to the difficulties observed with PHHs, human cell lines with varying metabolic expression profiles have served as models for hepatic drug metabolism and related toxicities. One, the HepG2 cell line, was isolated in 1975 from a 15-year-old boy with liver cancer, first described as hepatocellular carcinoma and later classified as a hepatoblastoma.29, 30, 31 Significant advantages of the HepG2 cell line include cost-effectiveness, wide availability, and simplicity of cell culture conditions compared to PHHs. The HepG2 line was originally observed to bioactivate the prodrug cyclophosphamide, a cytochrome P450 (CYP) CYP3A4 substrate, leading to genotoxic effects.³⁰ Despite this initial intriguing finding, the HepG2 cell line was later found to experience limited expression of many key drug metabolizing enzymes (DMEs) and transporters.^30,32 HepG2 cells have significantly lower mRNA expression of all CYP1, CYP2, and CYP3 subfamilies than PHHs.^33,34 Additionally, the HepG2 line demonstrates lower metabolic activity when incubated with common cytochrome P450 indicator substrates such as phenacetin, bupropion, diclofenac, midazolam (MDZ), and others.^35,36 owing to this, further research has been conducted to transfect/transduce HepG2 cells to increase CYP expression and activity in this cell line, with a specific focus on CYP3A4.37, 38, 39 The HepG2 line has also been used to study xenobiotic induction, a major component of drug-drug interactions (DDIs).^36,40

An alternative to HepG2 cells, the immortalized hepatic cell line Huh7, was originally established in 1982 from a highly differentiated hepatocellular carcinoma in a 57-year-old man.^41,42 While the Huh7 cell line has primarily been used to study the efficacy and metabolism of liver cancer drugs, it has also been considered a convenient experimental substitute for PHHs similar to HepG2 cells. Specifically, Huh7 cells have been useful in investigating drug disposition mediated by hepatic drug transporters and the interaction of drugs with multidrug resistance–associated proteins.^43,44 Additionally, other studies have explored their role in the metabolism of specific enzyme probe substrates, such as MDZ and the immunosuppressant tacrolimus, primarily through monitoring CYP3A4 metabolic activity.^40,45 It is important to note that Huh7 cells are homozygous for the CYP3A5∗3 null allele, which means that metabolism of MDZ and other CYP3A substrates will be driven only by CYP3A4.⁴⁶ Huh7 cells have been successfully employed in toxicity studies and are known to bioactivate certain compounds, such as acetaminophen (N-acetyl-p-aminophenol, or APAP) to produce the hepatotoxicant NAPQI.^47,48 While the DME activities of the HepG2 and Huh7 cell lines are lower than PHHs, they represent a stable, cost-effective alternative for certain drug metabolism studies.^48,49

Because of the limitations of PHHs, HepG2, and Huh7 cell lines, alternatives have been sought that incorporate both accuracy of primary hepatocytes with the reliability and ease of use of the immortalized cells. The HepaRG cell line originates from a hepatocholangiocarcinoma originally isolated from a female patient in 1999, representing a bipotential hepatic precursor.⁵⁰ Compared with the HepG2 and Huh7 cell lines, differentiated HepaRG cells display increased DME and drug transport functions, making them a more suitable model for drug metabolism and hepatotoxicity studies.⁵⁰ For CYP3A4 activity specifically, relatively high MDZ 1'-hydroxlyation and testosterone 6β-hydroxylation activities have been reported, often at similar levels to PHHs.^36,50 Furthermore, HepaRG cells display UDP-glucuronosyltransferase (UGT) and sulfotransferase activities comparable with or higher than PHHs.³⁵ When cultured with more than 20 distinct hepatotoxicants, the HepaRG cells had IC₅₀ or LD₅₀ values similar to that of PHHs in many cases.^35,51 Furthermore, HepaRGs express more uptake and efflux transporters compared with HepG2 cells, expanding their potential utility additionally to transport studies.⁵¹ The HepaRG cell line has a slightly more intensive culture protocol than HepG2 and Huh7 cell lines, with 2 weeks of culture in their undifferentiated bipotential state, and another 2 weeks of culture with DMSO to induce differentiation.^36,50 In general, the hepatic cell lines are easier to culture and more cost-effective, but have varying levels of phase I and II DME activity and drug transport when compared with PHHs, emphasizing how model selection plays a key role if immortalized cell lines are to be used in preclinical studies.

2.2. Human intestinal cell models

The intestine plays a key role in first-pass metabolism of xenobiotics before they reach the liver and ultimately systemic circulation. Primary enterocytes can be isolated directly from human jejunum or ileum and express a variety of DMEs such as CYP3A4, UGTs, and carboxylesterase 2.52, 53, 54 They closely mimic the in vivo intestinal environment; however, like primary hepatocytes, they are more difficult to culture, have a short lifespan, and suffer from low scalability.⁵⁵ Caco-2 cells are the most widespread model system to study the transport and absorption of drugs in the intestine. The Caco-2 cell line was originally isolated in the 1970s from a male patient with colon adenocarcinoma.^56,57 Caco-2 differentiated cells have high expression of many of the influx and efflux transporters found in the small intestine, such as P-gp, BCRP, OATPs, multidrug resistance–associated proteins, and OCTs.^58,59 High transporter expression makes the Caco-2 cell line appropriate for absorption and permeability studies; however, the lack of enzyme expression, namely CYP3A4, makes them a poor model for intestinal first-pass metabolism. There have been recent advances to overcome this issue, such as tetracycline-controllable CYP3A4 expression in Caco-2 cells and the screening of additional human colorectal cancer lines for high CYP3A4 expressers.60, 61, 62

2.3. Human kidney cell models

The kidney plays an important role in the elimination of hydrophilic drugs and is often affected by drug-induced kidney injury (DIKI), a major burden on the health care system.^63,64 Additionally, it has been known for some time that the kidney has the capacity to metabolize drugs, although not to the same extent as the liver.⁶⁵ Historically, primary renal proximal tubule epithelial cells (RPTECs) have been used to study kidney nephrotoxicity in vitro.^66,67 RPTECs are the most correlated with DIKI due to their high transporter expression, including OCT2 and OATP1, and have been shown to predict DIKI in a similar fashion to animal models.^68,69 Analogous to the other primary cells discussed in this review, the main drawback of RPTECs is that they are prone to de-differentiation and suffer from loss of transporter expression.⁷⁰ The HK-2 immortalized cell line, first reported in 1994, still remains a popular kidney toxicity model owing to the ease of cell culture.⁷¹ However, when compared against RPTECs or in vivo human data, HK-2 cells had varying responses to known nephrotoxicants, as quantified by ATP depletion indicating cell viability and relative expression of KIM-1, a biomarker for acute kidney injury.^72,73 RPTEC and HK-2 cells have varying expression of some DMEs, notably CYP2B6, UGT1A9, and UGT2B7 at the protein level.⁷⁴ The kidney is known to express CYP2B6, CYP3A5, UGT1A9, and UGT2B7, which could all play a minor role in systemic drug metabolism and clearance.⁷⁵

2.4. Coculture systems

The coculture of multiple cell types has been established to increase the complexity of these cell models and better represent the metabolic abilities of different organs or different cell types within an organ. Hepatocytes have been cultured with fibroblasts, hepatic stellate cells, and other nonparenchymal cells to better mimic the cellular diversity of the liver. An increase in liver functionality, especially CYP expression and activity, was noted for multiple coculture systems.76, 77, 78 Additionally, the inclusion of Kupffer cells (resident liver macrophages) allows for an evaluation of inflammatory responses during hepatotoxicity.^79,80 In addition to other hepatic cells, hepatocytes can also be cocultured with intestinal cells.^81,82 This has been effectively used in MPS systems, as discussed in the upcoming section. RPTECs have been cultured with other cell types, such as fibroblasts, to improve barrier integrity, transporter expression, and metabolism.⁸³

3. Key recent advances: novel cell culture models in drug metabolism and toxicity

More advanced cell culture systems, depicted in Fig. 2, have been recently used to improve upon weaknesses in traditional and immortalized cell culture techniques described earlier. In this section, we will cover novel cell culture models including human iPSC-derived primary-like cell monolayers and MPS/OoC model systems. A condensed summary describing the strengths and weaknesses of each model system is also provided in tabular form (Table 1).84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128

Fig. 2.

Novel cell culture systems to study drug metabolism and toxicity in vitro.

Table 1.

Summary of traditional and novel cellular and organoid models for drug metabolism and toxicity studies

System	Cell Types/Tissues	Advantages	Limitations	Literature Examples
Traditional 2D cell culture	Primary human cells (hepatocytes, enterocytes, renal proximal tubules, and epithelial cells)	Physiologically relevant in vitro system	Limited availability, cost, short lifespan and de-differentiation, and lack of 3D architecture	22,52, 53, 54,68,69
	Immortalized cell lines (HepG2, Huh7, HepaRG, Caco-2, and HK-2)	Low-cost, widely available, easy to culture	Lower DME expression, lack of 3D architecture	35,36,40,45,47,48,50,51,58,59,72,73
iPSC-derived models	Primary-like cells derived from human iPSC (hepatocyte-like cells, hiHeps, enterocyte-like cells, and small intestinal epithelial cells)	Renewable, available from a wide variety of donors, longevity in culture	Reproducibility, differentiation ability, full maturation, and scalability	84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94
MPS/OoC models	Microfluidic system with multiple human cell types (any listed above)	Dynamic microenvironment, multiorgan, media flow	Technically advanced, cost, scalability	53,77,95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117
3D organoids	iPSC-derived human hepatic organoids, intestinal stem cell–derived gut organoids, iPSC-derived kidney organoids	Self-organization, mimicry of in vivo physiology and 3D architecture, and enhanced maturity	Cost, complex culture and lack of standardized protocols, scalability, limited vascularization, innervation, and immune integration	53,61,65,111,118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128

3.1. Stem cell–derived monolayer systems

In the past several years, human iPSCs have been a popular choice as a starting material for production of primary-like cells by recapitulating primary tissues from a sustainable source. iPSCs can be differentiated into the cell type of choice by specific media and growth factors mimicking in vivo cellular development. iPSCs, first reprogrammed from fibroblasts, can be differentiated into both 2D monolayers, as discussed in this article, or into 3D organoids, as discussed later in this review. For the liver, the goal is to obtain differentiated, iPSC-derived hepatocyte-like cells (HLCs) that recapitulate drug metabolism activities of PHHs. Several groups have showed the utility of HLCs in drug metabolism, toxicity screening, and specific liver disease states.84, 85, 86, 87, 88 However, achieving full metabolic activity in HLCs is still a challenge, and no iPSC to HLC differentiation protocol to date has been able to yield HLCs that represent a full adult hepatic phenotype.¹²⁹ On the contrary, these HLCs embody a unique tool to evaluate drug metabolism and toxicity in certain special population groups, such as fetal/neonatal patients, where higher expression and activity of fetal/neonatal enzymes such as the CYP3A isoform CYP3A7 are observed.¹³⁰ hiHeps represent yet another hepatocyte model, directly reprogramming fibroblasts to HLCs and bypassing the stem cell stage.⁸⁹ hiHeps have shown some drug metabolizing enzymatic activities but are often still less active than primary hepatocytes and suffer from a more immature phenotype.^131,132

Similarly, intestinal cell models can be derived from human iPSCs. These iPSC-derived cells (enterocyte-like or small intestinal epithelial cells [SIECs]) have been probed for their absorption, transport, and first-pass metabolism capabilities through P-gp, BCRP, and CYP3A activities.90, 91, 92 Additionally, SIECs have been used to demonstrate gut toxicity mechanisms through treatment with drugs known to have intestinal cytotoxicity.⁹³ Alternatively, duodenal stem cells from donors can be used as an another starting material to produce enteroid monolayers. Enteroid monolayers have demonstrated their utility by being able to model a natural product-metformin transporter interaction commonly observed in diabetic patients.⁹⁴ For the kidney, iPSC-derived models are more common in the form of kidney organoids, rather than monolayers, and will be discussed later in the review. Overall, these iPSC-derived cell monolayers represent a promising alternative to primary human cells owing to their renewable nature and longevity in culture. Fibroblasts and/or iPSCs are available from a wide variety of donors with differing characteristics such as age, sex, disease status, ethnicity, and so on. Similar challenges present for all organ types, including reproducibility and differentiation efficiency, scalability, and full maturation; therefore further research is needed to address these bottlenecks.^129,133, 134, 135

3.2. Microphysiological systems/organ-on-a-chip models

MPS/OoC model systems are more complex than traditional 2D cell culture, combining a precisely engineered environment, microfluidic channels for media flow, and multiple cell types. MPS/OoCs have been developed for many organs, including the liver, intestine, and kidney for drug metabolism and toxicity purposes.

3.3. Liver-on-a-chip

Liver-on-a-chip (LoC) microfluidic devices, or MPS, can be seeded with PHHs, any of the previously discussed immortalized cell lines, or iPSC-derived HLCs to recreate the dynamic environment of the liver. LoC models have been used with a wide variety of drugs to study both their pharmacokinetics and pharmacodynamics, including the anticancer drugs 5-fluorourcil, capecitabine,⁹⁵ and irinotecan,⁹⁶ multiple drugs for nonalcoholic steatohepatitis treatment,⁹⁷ the Alzheimer disease drug donepezil,⁹⁸ MDZ, naloxone, and zidovudine.⁹⁶ Many studies have focused on CYP activity as the main contributing pathway to drug clearance, but some also include UGTs and aldehyde oxidase. In addition, LoC systems have been successfully used to recapitulate DDI, demonstrating the classic rifampicin-induced MDZ AUC changes out to 4 weeks, much longer than the 3–6 days of standard PHH 2D culture.⁹⁹ Further advanced characteristics of the liver have been assessed in LoC systems, such as oxygen zonation and its effect on metabolic function in hepatocytes.¹⁰⁰ Additionally, LoC systems have successfully recapitulated liver toxicity when dosed with known drug-induced liver injury (DILI) compounds, most commonly APAP to generate the toxic species NAPQI.100, 101, 102, 103 LoC systems have been mainly been developed with human cells, but it is important to note that other species can also be used to compare traditional preclinical animal models.¹⁰¹

3.4. Intestine and kidney

Intestine-on-a-chip (IoC) microfluidic devices can be seeded with 1 or more intestinal cell types to mimic the in vivo intestinal environment via directional culture media flow. IoC devices can be seeded with Caco-2 cells, iPSC-derived enterocyte-like cells or SIECs, ex vivo tissue, or organoids (as discussed below), which can expand their utility.^53,104, 105, 106 In general, various groups have found that IoC models more accurately reflect the in vivo intestinal environment, but they are costlier and more complicated to develop and/or scale up. The development of kidney-on-a-chip systems modeling the proximal tubule allows them to be seeded with a monolayer of RPTECs or HK-2 cells, with primary cells outperforming the cell lines in terms of barrier formation and transporter expression.107, 108, 109 Kidney-on-a-chip systems can be used as a HTS tool to evaluate drugs and preclinical candidates for nephrotoxicity and/or drug-transporter interactions.

3.5. Multiorgan systems

A major advantage of the OoC/MPS model is the ability to combine organ types, similar to the advances seen with 2D coculture systems.¹⁷ Especially important for drug metabolism, intestine-liver chips have been useful during the past few years to get a more global idea of drug disposition including absorption, first-pass metabolism, and hepatic metabolism along the gut-liver axis.^77,110, 111, 112, 113 Additional organ combinations include liver-kidney systems that have been developed from patient-derived recellularized kidney slices,¹¹⁴ gut-liver-placenta to study drug disposition in pregnant people,¹¹⁵ liver-breast to evaluate the pharmacokinetics-pharmacodynamics relationship of doxorubicin for cancer treatment,¹¹⁶ liver-brain to study the metabolism of glioblastoma drugs,¹¹⁷ and others.

4. Key recent advances: organoid models in drug metabolism and toxicity

The lack of accurate models for IVIVE has prompted the need for even more complex systems that include multiple cell types in a higher level of cellular and tissue organization. One of the most important advancements in this area has been the development of organoid systems.^6,18 Organoids were specifically produced in order to improve our ability to recreate some of the complex metabolic and transport patterns observed in tissues and organs.¹⁸ As the name implies, organoids more faithfully represent the functions of an entire organ than typical individual cell-type culture systems. In the simplest sense, organoids are self-contained, miniaturized, 3D cellular architectures grown in a laboratory that closely mimic the structure, organization, and functions of actual tissues and organs (Fig. 3). One major advantage over traditional 2D cell culture is that organoids can contain multiple highly specialized cell types, as well as the unique ECM that is important cellular growth and communication.^11,18 Another advantage of organoids is that they may be cultured for much longer periods than 2D single cell-type cultures, on the order of weeks in many cases, which also can improve the accuracy of IVIVE, particularly for low clearance drugs.^6,14,16

Fig. 3.

Common types of organoids and their uses in drug metabolism and toxicity studies.

In organoids, the cells and associated ECM self-organize into higher-level structures that can exhibit many of the key metabolic, structural, and functional features of their corresponding tissues or organs.^11,16 By enabling cells to grow and interact in a 3D space, these systems facilitate intricate cell-cell and cell-matrix interactions that are often absent or poorly represented in 2D systems.^7,11 For instance, spheroids develop an architecture characterized by complex cell-cell and ECM interactions, where cell-surface integrins and cadherins play vital roles in regulating adhesion, organization, and structural integrity among cells.^136,137 Organoids can contain any number of organ-specific cell types arranged in a spatial organization that mimics the architecture of the original tissue. This flexibility has allowed organoids to be produced for a number of organ types, including the brain,^138,139 the kidney,^69,140 intestine,^53,61,110 the liver,76, 77, 78^,141 and the heart.^23,142,143 These organoid systems can perform many of the specialized functions of organ-specific tissue types, such as uptake, metabolism, and transport, which make them ideal for studying ADMET in vitro.

4.1. Liver organoid systems

Given that the liver is the primary site of drug metabolism and detoxification in the body, liver organoids in particular have garnered a significant amount of attention as ideal test systems for preclinical ADMET studies. Human hepatic organoids (huHOs), typically derived from human iPSCs, have been designed to mimic the physical and biochemical features of in vivo human liver models with advanced phenotypic characteristics and functionality.¹¹⁸ These organoids were demonstrated to successfully replicate the liver’s complex physiological structure, various functions, and the diversity of cell types present, including hepatocytes, cholangiocytes, and Kupffer cells.^124,144 Functional validation of huHOs has included the demonstration of glycogen storage,¹¹⁹ the secretion of albumin and bile acids,^119,120 and the assessment of DMEs, including CYP enzyme activity.119, 120, 121 Differentiated organoids consistently show significantly increased albumin and bile acid secretion, along with higher CYP3A4 activity, than their undifferentiated counterparts.^119,121 Furthermore, they exhibit functional polarization, characterized by the presence of bile canaliculi that actively excrete bile acids, a critical feature for liver function.¹¹⁹ The expression of CYP enzymes in human iPSC-derived hepatic models is directly correlated with the maturity of the liver cells within the model.¹²² Liver organoids generally achieve a more mature phenotype than 2D iPSC-derived HLCs, which often display fetal-like CYP expression profiles and lack sufficient CYP-mediated hepatic metabolism, limiting their utility for adult drug metabolism studies.¹²² huHOs have also been shown to serve as physiologically relevant in vitro platforms for the evaluation of hepatotoxicity.¹²³ Additionally, they possess the remarkable ability to effectively distinguish between hepatotoxic and nonhepatotoxic substances.¹¹⁹ For example, known DILI agents such as ketoconazole, troglitazone, and tolcapone exhibit significantly lower TC₅₀ values in organoid assays than nonhepatotoxic compounds such as sucrose, ascorbic acid, and biotin.¹¹⁹

Another advantage is that traditional biomarkers used for hepatotoxicity assessment in vivo may also be used in organoids, including liver failure markers such as albumin, alanine aminotransferase, and aspartate aminotransferase, as well as various cell viability assays (eg, ATP assay, MTT assay, and lactate dehydrogenase release assay).¹²³ Beyond these, novel amino acid (AA)-based biomarkers, including specific AAs such as aspartic acid, arginine, glutamine, and phenylalanine, and their ratios, have shown reliable alterations in huHO media upon drug treatment. The changes observed in these biomarkers can enable the differentiation of hepatotoxic from nonhepatotoxic drugs.¹²³ A significant advantage of these AA-based markers is their nondestructive nature, allowing for analysis directly from the media without damaging the organoids, and their potential for standardization.¹²³ Liver organoids have demonstrated high accuracy in predicting liver toxicity, with studies showing reliable outcomes when screening a large panel of marketed drugs, including 206 known DILI compounds.^27,79,125 Furthermore, these models can be used to study specific liver diseases, such as nonalcoholic fatty liver disease and nonalcoholic steatohepatitis, by exhibiting lipid accumulation, inflammation, and fibrosis when exposed to free fatty acids.¹²⁵

The observation that CYP expression in human pluripotent stem cell–derived liver models is directly correlated with liver maturity signifies a crucial link between cellular differentiation state and functional relevance for drug metabolism studies.¹²² This explains why organoids, being more mature than 2D iPSC-derived HLCs, are superior models for DILI and drug metabolism.¹²² The capacity of these models to accurately distinguish between hepatotoxic and nonhepatotoxic substances and their high predictive accuracy in screening for liver toxicity are direct consequences of this enhanced maturity and functional mimicry. Therefore, continued efforts to drive organoid maturation are paramount for further improving their predictive accuracy in drug metabolism and toxicity studies, particularly for CYP-mediated reactions, solidifying their role as indispensable tools in preclinical drug development.

4.2. Intestinal organoid systems

The gastrointestinal tract serves as a primary organ for drug absorption, particularly for orally administered medications. Beyond absorption, the gut plays a complex and multifaceted role in metabolism, detoxification, and immune responses.145, 146, 147, 148 Gut organoids, derived from intestinal stem cells, possess the remarkable ability to differentiate into various specialized cell types found within the intestinal epithelium, including absorptive enterocytes, goblet cells, and Paneth cells.^53,61 These organoids successfully recapitulate the complex architectural features and diverse functions of the in vivo intestinal epithelium.¹²⁶ They are invaluable tools for investigating fundamental processes such as nutrient metabolism, drug absorption, drug metabolism, and the intricate interactions between the host and its microbiome.^126,145

Gut organoids can effectively simulate drug metabolism in a manner analogous to the human intestine. Studies have observed the induction of key DMEs and transporters, such as CYP3A4 and ABCB1 gene expression, indicating their capacity to metabolize drugs similarly to the human gut.¹²⁶ These models are well-suited for preclinical toxicology and pharmacokinetic studies, providing insights into how drugs are processed and distributed within the body. For instance, gut organoids have been successfully used to assess the efficacy of antiviral drugs, such as remdesivir targeting SARS-CoV-2, and have demonstrated high accuracy (90%) in predicting drug-induced gastrointestinal toxicity for a reference set of 31 drugs, outperforming traditional rodent models.¹²⁶

Intestinal organoids have also shown promise as models to study gut toxicity as the intestinal epithelium functions as the body’s first line of defense against a multitude of ingested toxins.¹²⁶ Exposure of gut organoids to various xenobiotics, encompassing both environmental toxins and pharmaceutical compounds, may reveal potential damage to the digestive system and beyond. This includes compromised nutrient uptake and a weakening of the crucial gut barrier function.¹²⁶ Examples of such studies include observations of reduced cell viability, impaired differentiation, induction of apoptosis, dysfunction in mucus production, and damage to the epithelial barrier when exposed to heavy metals like cadmium and lead.¹²⁶ Furthermore, gut organoids have been instrumental in studying genotoxicity, as demonstrated by the upregulation of xenobiotic-metabolizing enzyme genes (eg, CYP1A1 and NQO1) upon exposure to carcinogens like benzo[a]pyrene, and in assessing the impact of various dietary compounds, nanoparticles, and food additives.^53,111,126

The explicit connection made between gut and liver organoids, stating that the liver is the primary site of detoxification and metabolism of xenobiotics, usually routed from the gut, highlights the critical concept of first-pass metabolism and the gut-liver axis.¹²⁶ This physiological relationship is fundamental to understanding the systemic bioavailability of orally administered drugs and their potential for systemic toxicity. The ability to model this complex interaction, particularly when gut organoids are integrated into multiorgan-on-a-chip MPS systems, represents a significant advancement over single-organ models.⁵³ Gut organoids are therefore not only valuable for investigating localized intestinal effects but are also indispensable components for accurately modeling systemic drug absorption and first-pass metabolism, especially when integrated with liver models, thereby providing a more complete picture of oral drug pharmacokinetics.

4.3. Kidney organoid systems

The kidney is an indispensable organ for the elimination of hydrophilic molecule from the body, including therapeutic drugs and their metabolites.^65,127 As mentioned previously, drug nephrotoxicity, or DIKI, represents a common health care challenge and a significant limiting factor during drug development, accounting for 7%–20% of all drug-related toxicities.¹²⁸ Kidney organoids, derived from human iPSCs, are engineered to contain epithelial nephron-like structures, including podocytes, proximal tubules, loops of Henle, and distal nephrons.^53,69,140 These structures are arranged in an organized and continuous manner that closely resembles the in vivo nephron architecture.¹²⁸ Such models are capable of recapitulating various kidney disorders, such as polycystic kidney disease and acute kidney injury.¹²⁸ In regards to drug and xenobiotic transport, the epithelial cells of kidney tubules express a diverse array of membrane carriers and transporters.¹²⁷ These transporters are critical for drug elimination, can influence drug nephrotoxicity and DDI, and may also serve as direct drug targets.¹²⁷ Kidney organoids have been developed that can specifically express key renal drug transporters, including OAT1, OAT3, and OCT2, predominantly located in the proximal tubules.^69,127 The plasma membrane monoamine transporter (PMAT) is also expressed in podocytes.¹⁴⁹ OAT1 and OAT3 are responsible for the secretion of negatively charged drugs into tubular cells, while OCT2 facilitates the uptake of positively charged drugs.¹⁵⁰ Notably, the expression levels of OAT1/3 and OCT2 in kidney organoids are significantly higher than those in traditional human proximal tubular cell lines, rendering organoids more physiologically relevant for studying drug transport.¹⁵⁰ Kidney organoids can serve as valuable tools for nephrotoxicity assessment due to their complex multicellular context and physiologically relevant transporter expression.¹²⁷ Drug-induced nephrotoxicity has been examined in kidney organoids and well recapitulates the whole-animal phenotype. Drug-induced nephrotoxicity can manifest through several mechanisms: proximal tubular injury and acute tubular necrosis, often dose dependent and related to apical contact or basolateral secretion of drugs; tubular obstruction caused by drug crystals or casts; or acute tubulointerstitial nephritis, which is typically a dose-independent, T cell–mediated hypersensitivity reaction.¹⁵¹

4.4. Cardiac organoid systems

Off-target cardiotoxicity is a major red flag during the drug development process. Recently, cardiac organoids derived from iPSCs have been used to demonstrate cardiotoxicity, especially with doxorubicin. Cardiotoxic damage in the form of apoptosis, inflammation, fibrosis, and contractile impairment have all been assessed in cardiac organoids and generally outperform traditional 2D cell cultures or animal models.^23,142,143

5. Current challenges and knowledge gaps: integrating novel cell culture systems into drug development

5.1. Comparison of in vitro models with animal models

The consistent observation of high failure rates in clinical trials, largely attributed to a lack of efficacy and safety, underscores a fundamental problem: the predictive validity of existing preclinical animal models. Despite their long-standing designation as the gold standard for preclinical validation, animal models frequently fail to accurately forecast drug metabolism, efficacy, and toxicity in humans. This issue extends beyond a purely scientific challenge because it also imposes substantial economic burdens on pharmaceutical companies through increased research and development costs and drug withdrawals. In addition, the extensive reliance on animal testing presents a significant ethical dilemma. The lack of correlation to humans is primarily due to significant species-specific differences in anatomy, physiological function, morphology, drug metabolic pathways, and immune responses.152, 153, 154 For example, humans can exhibit vastly different sensitivities to certain drugs compared with animals; rats and dogs, for instance, may tolerate 4.5- to 100-fold higher concentrations of some compounds than humans.¹⁵⁵ This interspecies discrepancy contributes to a high percentage of drugs (ranging from 40% to over 90%) that appear safe and effective in animal studies but subsequently fail in human clinical trials due to unforeseen safety or efficacy issues.¹⁵⁶ Conversely, certain compounds are known to exhibit toxicity in certain preferred preclinical animal species, but not in humans (eg, theobromine in dogs).^155,157,158

The increasing acceptance of MPS systems and organoids by regulatory bodies, such as the FDA, as alternatives to animal testing signals a growing recognition of these limitations and a proactive endorsement of more human-relevant in vitro models.¹⁷ This regulatory shift suggests a future where data derived from these advanced in vitro systems will play an increasingly pivotal role in drug approval processes. As noted at the outset, the FDA intends to shift to NAMs for all preclinical IND testing by 2030.²⁰ However, many hurdles need to be overcome before the NAMs will replace all preclinical animal testing. In this regard, recent studies have demonstrated that 3D organoid and MPS systems perform favorably when compared with traditional 2D platforms.^40,159,160

5.2. Current challenges in using novel systems

Despite their immense promise, the widespread adoption of 3D organoid cell culture techniques faces several significant challenges.⁶ A primary hurdle is the lack of consensus on optimal protocols for culturing, differentiating, and maintaining 3D cell cultures. This variability can lead to inconsistencies in results across different research groups.⁷ If experimental results vary significantly between laboratories owing to nonstandardized protocols, the fundamental principles of scientific rigor, that is, reliability and reproducibility, will be compromised. Developing standardized protocols is therefore essential, particularly in regards to FDA IND submissions. It is likely that consensus on standardization of methodologies will eventually be reached through FDA guidance and policy white papers, following a similar example as the Chinese Society of Biotechnology and their recent standards and guidelines on LoC models.¹⁶¹ An additional issue, not solely limited to 3D models but perhaps more apparent in these systems, is data complexity. The intricate nature of 3D cell cultures generates vast amounts of complex data, making its acquisition and analysis particularly challenging. This necessitates the development of sophisticated analytical tools, including advanced computational models and image analysis algorithms. Here, the careful integration of novel AI algorithms will be essential for sorting and analyzing the large datasets that will be produced with these new methodologies. This highlights the imperative for interdisciplinary collaboration among biologists, engineers, and data scientists to develop these new tools.

While these in vitro systems are significantly cheaper than animal models, they still require a significant cost and time commitment. The specialized equipment, reagents, and longer culture durations required for 3D cell cultures make them inherently more expensive and time consuming than their 2D counterparts, which may offer limited data but in a much faster turnaround time.⁷ Efforts to optimize production processes are needed to reduce the overall cost and time to initial data acquisition. Similarly, scalability and integration with existing platforms, such as HTS, also remain roadblocks to more widespread adoption of these novel methodologies. Achieving industrial-scale production and seamlessly integrating 3D cultures into existing HTS pipelines and workflows are still significant obstacles.⁷ There are also technical obstacles that need to be addressed in order for development to proceed, including adequate nutrient supply and oxygen diffusion in long-term 3D organoid cultures. As 3D organoids increase in size, their inner cells can suffer from a lack of nutrients and accumulation of waste products due to diffusion limitations. This negatively impacts their growth and functionality, potentially leading to necrotic cores.¹⁶² Developing methods to increase the circulation of culture medium is crucial to address this issue.¹⁶² Finally, the bioactivity and potential toxicity of the materials used to construct 3D cultures are also critical factors that can influence experimental success.^9,10

6. Perspective on future directions

The field of novel cell and organoid systems is continuously evolving, driven by the imperative to create more predictive, efficient, and ethically sound models for drug discovery and toxicology. Several key trends are currently shaping the future landscape and are likely to have a significant impact in the near future.

6.1. Integration of AI and ML

The integration of AI and ML with OoC platforms is emerging as a transformative synergy in drug discovery and development.¹⁵ AI possesses the capability to digest vast, complex datasets and intricate biological networks, thereby significantly impacting our ability to analyze the large and rich datasets generated by organoid and MPS systems.¹⁶³ When ML algorithms are applied to the rich data generated by OoC experiments, they enable more sophisticated predictive modeling of tissue responses to various drug candidates. This capability is poised to revolutionize drug screening methodologies by offering more efficient and cost-effective alternatives to traditional pipelines.164, 165, 166

AI and ML can further enhance preclinical studies by providing advanced analytical capabilities to address complex factors such as environmental influences, lifestyle variations, genetic diversity, dynamics of nutrient supply, and intricate drug metabolism pathways.^15,165 This integration streamlines data analysis processes, accelerates drug screening, facilitates real-time monitoring of cellular responses, and enhances the overall capabilities for disease modeling. The substantial increase in data generated by integrated OoC systems, coupled with the growing demand for robust data management and sophisticated analyses, indicates that the complexity and sheer volume of data produced by these novel in vitro models will become a significant bottleneck.¹⁶⁵ This makes AI and ML not merely an enhancement but a necessity for extracting meaningful biological insights and maximizing the full potential of these advanced platforms. This synergy is essential for unlocking the full predictive power of these advanced in vitro models and accelerating the pace of drug development.

6.2. Advancements in vascularization, innervation, and immune components

A significant limitation of current organoid models is the absence of fully functional vascularization, innervation, and integrated immune cell components. These elements are crucial for providing adequate nutrition, transmitting physiological cues, and mediating critical interactions, and their absence can undermine the validity of organoids as truly representative physiological or pathological models.¹⁶⁷

The development of a robust vascular system is paramount for preventing necrotic cores in larger organoids.¹⁶⁷ Current strategies for achieving vascularization include in vivo engraftment into immune-deficient hosts, coculturing organoids with endothelial cells or mesodermal progenitor cells, and using gene editing techniques to induce an endothelial fate within the organoid structure.¹⁶⁷ Advances in bioengineering, particularly in 3D bioprinting, are expected to significantly enhance the scalability and robustness of vascularized organoid production.^168,169 Some advanced multiorgan-on-a-chip systems are already integrating endothelial barriers and circulating vascular flow to more closely resemble systemic drug circulation and distribution within the body.¹⁷

Endowing organoids with functional immune cells also remains a major challenge, and standardized methods for this integration are largely undeveloped.¹⁶⁷ The most common approach involves coculturing organoids with specific immune cells, which can be either immortalized cell lines or autologous immune cells derived from patients.¹⁶⁷ Furthermore, human immune organoids, such as lymph node or spleen organoids, and blood-on-a-chip systems incorporating circulating immune cells are currently being developed. These models aim to enable the testing of cytokine release, T cell activation, and other immunotoxicities, which are particularly relevant for the development and safety assessment of monoclonal antibody therapies.¹⁵⁶

While less extensively detailed in the current literature, the broader goal of mimicking the full in vivo microenvironment implicitly necessitates the integration of innervation, especially for organs with significant neural control, such as the gut and brain. Similarly, complete vascularization is needed to truly mimic complete organ systems. The development of vascularized multiorgan chips represents a direct and significant step in this direction, transitioning from the study of isolated organ function to the comprehensive analysis of systemic integration. Future advancements in novel in vitro models will heavily focus on integrating vascularization, innervation, and immune components, as these are essential for truly comprehensive and predictive drug metabolism and toxicity assessment, particularly for understanding long-term and immune-mediated drug effects.

6.3. Personalized medicine and precision toxicology

Novel cell and organoid systems hold immense potential for revolutionizing personalized medicine and precision toxicology.¹⁶² For example, patient-derived organoids are capable of accurately reflecting the unique genetic and phenotypic traits of individual patients. This enables highly targeted drug screening, particularly in the field of precision oncology, where treatments can be tailored to specific tumor characteristics.¹⁶² Another important area where these models may make serious advancement is in examining interindividual differences. Many of the advanced systems described in this article provide an unprecedented opportunity to understand the significant interindividual heterogeneity in drug responses, in particular regarding metabolism and toxicity. This understanding is crucial for tailoring therapeutic regimens to individual patients, moving away from a 1-size-fits-all approach.¹⁶² For instance, by employing CRISPER-Cas9 gene editing technology, iPSC-derived HLCs can be used to explore the role of genetic diversity in DILI, potentially leading to personalized drug administration strategies that account for an individual’s genetic predisposition.¹⁵⁵ The ultimate promise of novel cell and organoid systems therefore lies in their capacity to revolutionize personalized medicine and precision toxicology, enabling the development of tailored therapies that maximize efficacy while minimizing adverse effects by accounting for individual patient variability.

7. Conclusion

The landscape of drug discovery and toxicology is undergoing a profound transformation, driven in part by the urgent need to overcome some of the inherent limitations of traditional 2D cell cultures and animal models. These conventional approaches, while still useful, suffer from a significant predictive validity gap that contributes to high drug attrition rates, substantial economic burdens, and ethical concerns. Novel cell and organoid systems, including advanced 3D cell cultures, self-organizing organoids, iPSC-derived models, and sophisticated OoC/MPS technologies, are emerging as powerful solutions to these challenges. These advanced in vitro platforms offer enhanced physiological relevance, improved predictive accuracy, and compelling ethical advantages. They achieve this by meticulously mimicking the complex in vivo microenvironment, enabling more accurate assessments of drug efficacy, metabolism, and toxicity.

The synergistic integration of these models into multiorgan-on-a-chip systems represents a revolutionary step toward recapitulating systemic drug fate and interorgan communication. These platforms facilitate comprehensive pharmacokinetics and pharmacodynamics studies, enabling the early identification of secondary drug toxicity and a more accurate prediction of in vivo drug behavior. The ability to generate patient-specific organoids and iPSC-derived cells is a cornerstone for advancing personalized medicine and precision toxicology, allowing for the development of tailored therapies that account for individual patient variability, thereby maximizing efficacy and minimizing adverse effects.

Despite these transformative advancements, significant challenges remain. The lack of standardized protocols for 3D cell cultures and organoids continues to pose hurdles for reproducibility and widespread industrial adoption. The absence of complete vascularization, innervation, and integrated immune components in many current organoid models represents a missing link in achieving full physiological completeness, limiting their utility for long-term and immune-mediated drug effects. Furthermore, the complexity of data generated by these advanced systems necessitates the continued integration of AI and ML for robust analysis and interpretation.

The future success of these novel cell and organoid systems in revolutionizing pharmaceutical development is contingent upon addressing these challenges through sustained research and interdisciplinary collaboration among biologists, engineers, pharmacologists, and data scientists. By doing so, these innovative platforms hold the promise of leading to the development of safer, more effective, and more efficiently produced pharmaceuticals for patients worldwide.

Conflict of interest

Jed N. Lampe reports financial support was provided by NIH National Institute of Allergy and Infectious Diseases and is editorial board member of Drug Metabolism and Disposition. The other authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article.