Concise Review: Human Pluripotent Stem Cell-Based Models for Cardiac and Hepatic Toxicity Assessment§


  • Author contributions: P.S. and P.B.: conception and design, manuscript writing, final approval of manuscript.

  • First published online in STEM CELLS EXPRESS March 23, 2011.

  • §

    Disclosure of potential conflicts of interest is found at the end of this article.


Considering the costs associated with drug development, there are billions of dollars to be saved by reducing late-stage attrition in the pharmaceutical industries. Reports on the use of human pluripotent stem cells (hPSCs) and their functional derivatives in applications for safety assessment of drugs have begun to appear in the scientific literature. These reports are encouraging and fuel further developments of improved human cellular models that may increase the clinical relevance and reduce the need of experimental animals in preclinical drug discovery. However, a few factors still limit the general and wide-spread industry implementation of these new stem cell-based models, including cost of manufacture, level of functionality of the differentiated cells, assay validation, verification of human relevance, and benchmarking to conventional models. This review discusses the emerging field of hPSC-based models for drug discovery and development with a focus on cardiac and hepatic toxicity testing and how these approaches may improve current applications used in the pharmaceutical industry. Although much research remains to make hPSC-based models mainstream tools in the industry, importantly, this review highlights currently available opportunities. In addition, a forward looking discussion on novel applications using tissue preparations generated from hPSCs illustrates the opportunities to create complex models in vitro with the aim of simulating the systemic response of a drug in vivo. STEM CELLS 2011;29:744–748


The average time to develop a new drug is between 10 and 15 years and the associated costs are at least in the magnitude of one billion USD. Notably, safety and efficacy issues are the two major reasons for late-stage attrition during drug development [1]. To bring more effective and less costly drugs to the market in the future, improvements of these two aspects are essential. With regard to the safety issues, success in late-stage development should be increased by frontloading of predictive toxicology assays that can assess potential safety concerns with high accuracy early in the drug development process. Two criteria should be met: rapid throughput and high predictive rate with few false-positive results. Establishing rapid safety screening requires automation and high-throughput format that permit a large number of compounds to be tested in a cost-effective manner. Beyond the scale of analysis, any toxicology screen in preclinical drug discovery needs to be predictive, and a high true positive rate and a low false-positive rate are required. Cell-based assays have the capacity to fulfill these criteria as they are amenable to high-throughput screening. However, the biological components of such systems (i.e., the cells) need to be of high and consistent quality to ensure accurate data output. The access to clinically relevant human cells displaying phenotypes closely related to their in vivo counterparts would clearly be advantageous. With the advent of human pluripotent stem cell (hPSC) technology [2, 3], the generation of such cells from an inexhaustible source has become a tangible reality. In this review, we will discuss the current efforts in using hPSCs for applications in pharmaceutical cardiac and hepatic toxicity testing. These examples are brought forth to give the reader a sense of what is already possible, and what may be a reality in the short-term. Although the discussion in this review focuses on the pharmaceutical industry, the chemical and cosmetic industries are also facing toxicological challenges that can be addressed using hPSC technologies.


The withdrawal of the nonsteroidal anti-inflammatory drug Vioxx from the market by Merck in 2004 due to cardiovascular safety concerns illustrates the tremendous cost in lost revenues and patient litigations caused by late-stage candidate failure and underscores the need for better and more predictive models [4]. The annual sale of Vioxx was 2.5 billion USD before the recall. As a result of the announcement, the Merck stock tumbled by one-third and, in 2007, a 4.85-billion USD agreement was entered to resolve Vioxx lawsuits.

Cardiotoxicity is a general term that describes myocyte damage leading to decreased cardiac function and may include disruption of electrical conduction through the heart causing arrhythmias. The electrophysiology of the heart can be perturbed by direct interactions of therapeutics with specific ion channels expressed by the cardiac myocytes that can disrupt normal ion conduction through these channels [5]. Approved therapeutics and potential therapeutics can and often do have unintended effects on potassium currents that may cause QT-prolongation and potentially fatal arrhythmias [6, 7]. In addition, other compounds may cause cardiac muscle damage without interacting with ion-channels. For instance, anticancer treatments are commonly associated with cardiotoxic events, although the mechanisms of toxicity vary widely among different chemotherapeutics and include inflammation, mitochondrial dysfunction, cardiomyocyte apoptosis, DNA damage, and generation of reactive oxygen species [8]. The experimental models to assess cardiotoxicity currently used in the pharmaceutical industries include transfected cell lines, primary animal cardiomyocytes, tissue and whole organ models, as well as small and large animal models [9]. Notably, none of these models are based on human material, and it is not unusual that the first time a new compound under development actually comes in contact with human cardiac cells or tissue is during the first clinical trial. However, this situation may be changed in the future because alternative renewable sources of human cardiomyocytes derived from hPSC lines are becoming available (Fig. 1). The goal of hPSC development for toxicity testing is to identify these risks to human tissue before the drugs reach the market.

Figure 1.

Frontloading of human relevant toxicity testing in preclinical drug discovery. The creation of specialized functional cells from human induced pluripotent stem cells (hiPSCs) or human embryonic stem cells (hESCs) in large volumes offers the opportunity to implement high throughput cell-based assays and to screen for possible side effects of new chemical entities, early in the drug development process. Normal hESCs have the advantage of representing cell populations that have a nonmodified genome and as such represent a “clean” system. On the other hand, hiPSCs can be created from patients carrying specific diseases or from individuals with selected genetic backgrounds, thus allowing drug testing on healthy and disease human cell material as well as material that is representative of so-called “responders” and “nonresponders” within a given population. Simple models in two-dimensions, typically in a plate-based format, can fit in already existing platforms for high throughput and high content screening. More advanced systems can be created in three-dimensions, making use of bioreactor settings in which well-controlled tissue-like models recapitulate an “in vivo like” environment. Such models may potentially complete with tissue-based models and reduce the need for experimental animals. By frontloading the discovery and development process with clinically relevant toxicology tests, it is anticipated that late-stage attrition will be decreased. Abbreviations: hESC, human embryonic stem cell; hiPSC, human induced pluripotent stem cell.

The derivation of cardiomyocytes from hPSC has now been reported by many independent investigators and a variety of strategies and protocols have been published (reviewed in [10]). hPSC-derived cardiomyocytes display gene expression and microRNA expression profiles, ion channel functionality, ultrastructures, and pharmacological responses sharing similarities with an embryonic/fetal cardiac phenotype [11]. However, further maturation of the cells has also been reported after long-term culture in vitro [12, 13]. Typically, as part of the functional evaluation of hPSC-derived cardiomyocytes, the cells are subjected to dosed exposure of a subset of drugs with known and commonly accepted arrhythmogenic properties. For the purpose of demonstrating relevant functionality of the differentiated hPSC-derived cardiomyocytes, several studies have used techniques to interrogate the cells' electrophysiology to illustrate characteristic drug responses [11]. Such results are encouraging, but at the same time appear somewhat scattered due to the low number of compounds used in each study and the general differences in experimental setups. Recently, an assay system for drug-induced QT interval prolongation using hPSC-derived cardiomyocytes and microelectrode arrays was reported [14]. Most of the selected drugs in this study were previously associated with QT prolongation and/or Torsade de Pointes in humans (e.g., Quinidine, D,L-sotalol, cisapride, and terfenadine) but also negative controls (ketoconazole and verapamil) were included to demonstrate specificity. Importantly, the authors reported a detailed dose-response analysis (in the range of patient serum levels) in which expected effects on QT-interval overlapped with prolonged field potential duration. This study provides a good starting point, but to obtain a higher level of confidence in this system and to gain industry acceptance, a larger collection of drugs need to be evaluated, and the cells produced under standard operating procedures governed by an industry standard quality assurance and quality control system.

As an alternative to the comparison of the drug effects with published results from other laboratories, two recent studies have demonstrated concordance between hPSC-derived cardiomyocytes and conventional, well validated, rabbit and canine ex vivo Purkinje fiber models, which are commonly used as follow-up assays [15, 16]. Using microelectrodes to measure the transmembrane action potential of hPSC-derived cardiomyocytes, Jonsson et al. [15] determined parameters such as reverse use dependence, triangulation of the action potential, and short-term variability of repolarization to assess drug-induced arrhythmic events. Importantly, for the comparison with the Purkinje fiber model, all data collection for this study was performed at the same laboratory, using the same experimental setup, and under the same conditions. The results demonstrated that rabbit Purkinje fibers and ventricular-like hPSC-derived cardiomyocytes reacted in a similar way with regards to the incidence of early after-depolarization, increased triangulation, and short-term variability of repolarization in response to the human ether-à-go-go–related gene channel blocking compound E-4031. In a separate study, Peng et al. [16] demonstrated that hPSC-derived cardiomyocytes used in an action potential assay provided excellent pharmacological sensitivity when compared with rabbit and canine Purkinje fibers. In addition, the use of hPSC-derived cardiomyocytes resulted in reduced compound consumption and cost and time savings when compared with conventional Purkinje fiber assays. They also showed that hPSC-derived cardiomyocytes are good detectors of proarrhythmic events and that early after-depolarizations were induced by the reference compounds terfenadine, sotalol, cisapride, and E-4031. Although hPSC-derived cardiomyocytes clearly do not represent a complete organism, taken together, these studies suggest that the use of such in vitro generated cells can contribute to a reduction of the number of experimental animals needed for tissue extraction for safety pharmacology applications.

As indicated above, it is known that the compounds that do not interfere with ion channel functionality can also cause cardiotoxic insults [8]. hPSC-derived cardiomyocytes appear well suited to study such effects in vitro. The assessment of hPSC-derived cardiomyocytes for use in doxorubicin-induced toxicity testing was recently reported [17]. This study used a testing strategy based on two clinically decisive biomarkers of cardiac damage, which are considered sensitive indicators for drug-induced toxicity. Importantly, hPSC-derived cardiomyocytes were shown to release detectable levels of cardiac troponin T and fatty acid binding protein 3 in a dose-dependent manner in response to doxorubicin [17]. Based on the availability of very sensitive and rapid analysis techniques for these biomarkers, the assay lends itself well to miniaturization and high through-put formats. In addition, as the measurements are performed using only a small sample of conditioned medium, it would be possible to analyze additional biomarkers and to combine the assay with other read-outs, including electrophysiology and/or high content analysis. To validate this testing strategy, a large collection of drugs with known effects in human needs to be screened and classified accurately. This platform may prove useful for dissecting the molecular mechanisms of cardiac side effects observed in response to drug molecules.


The term hepatic toxicity is in general used in the context of chemical induced liver damage. Based on the fundamental role of the liver in transforming and clearing xenobiotics, this organ is constantly exposed to a variety of agents which directly or indirectly may cause hepatic injury. The mechanisms by which drugs damage the liver are variable, complex, and often poorly understood but may include: bile acid-induced liver cell injury during cholestasis, mitochondrial dysfunction, cell damage by oxidative stress, and reactive metabolite formation.

Until now, research and industrial development of chemical substances have been dependent on animal experimentation. Rodent bioassays are frequently used in various studies designed to predict human liver injury. Obvious shortcomings of this strategy are high costs, low throughput, and limited human relevance. During the last decades, huge efforts to find and establish alternative human cell-systems have been assigned to this field. At present, primary human hepatocytes constitute a common model for in vitro safety assessment. Unfortunately, key enzymes involved in metabolizing known and potential therapeutics, as well as many transporter functions, are rapidly lost when these cells are kept in culture [18]. Moreover, many human liver cell lines, which have been validated as alternatives, are to a great extent lacking expression of the important functional enzyme systems [19]. Accordingly, there is an urgent need for new model systems that better mimic human liver cells and that are able to predict effects of candidate molecules in the development of new drugs or chemicals.

Both human embryonic stem cells (hESCs) and human induced pluripotent stem cells (hiPSCs) are considered as valuable sources for predictable human hepatotoxicity models [2, 3]. Pluripotent stem cells can be specified along the hepatic lineage in large quantities, which provides significant advantages over their adult counterparts [20]. A number of studies have demonstrated the feasibility of differentiating hPSCs to hepatocytes [21–23]. Although each group has used slightly different approaches, the resulting hepatocyte-like cells show many common hepatic features measured at gene, protein, and functional level. Even though there are now an increasing number of reports showing that homogeneous populations of hPSC-derived hepatocytes can be generated, there is still a need for improved differentiation and maturation protocols in this field. Interestingly, by using a chemical biology approach, Hay et al. [23] recently identified a polyurethane matrix that enhanced hPSC-derived hepatocyte functionality and long-term growth. This study underscores the critical importance of identifying optimal substrates for directing and sustaining specialized cellular phenotypes.

Sourcing of different hESC lines can address the need for genetic diversity in hepatotoxicity testing, but the recent advancements in hiPSC research will clearly advance the field even further [24, 25]. Access to cell lines from donors with attractive genetic profiles will permit design of more specific in vitro tests [26, 27]. A thrilling example is to generate hiPSC-derived hepatocytes from the subpopulations of patients affected by drug-induced liver injury after medical treatment. Such model systems have the potential to identify problematic drug-induced liver injury-causing compounds early in the drug discovery process, a selection not possible by using any experimental system available today.

The implementation of hPSC-derived hepatocytes for toxicity testing is still in its infancy. One major reason for this appears to be that few organizations are capable of producing enough cells at sufficient quality to run large well-controlled studies. However, to our knowledge, there are several toxicity studies recently finalized or ongoing. We have participated as the cell manufacturer in a major international program financed by the European Commission (acronym: Carcinogenomics), where multiple hepatotoxic compounds have been assayed in a hESC-derived liver model in 96-well plate format. From this study, it is clear that hPSC-derived hepatocytes have the potential to predict toxic responses and to group compounds according to their mechanism of toxicity (Yildirimman et al., manuscript in preparation). Current studies, and upcoming efforts like the European Commission and European Federation of Pharmaceutical Industries and Associations cofinanced program “Innovative Medicines Initiative on Drug-Induced Liver Injury” (, will most likely catalyze the development of new hPSC-derived assays in the coming years. To bring this field into the next level of validated studies, there is a need for additional large interdisciplinary projects involving industrial hPSC-manufacturers, toxicologists, biostatisticians, as well as pharmaceutical companies and regulatory bodies.

Ongoing efforts using hPSC-derived hepatic cells in safety assessment studies are promising, and assays based on such cells have the potential for high-throughput combined with predictive power. However, it is unlikely that all types of liver toxicity will be efficiently traced using monolayer cultures of pure hepatocytes. It is obvious that two-dimensional (2D) cultures of hepatocytes lack the physiological integration with other important cells and systems present within the liver that are required to amplify the initial toxicological lesion into overall organ toxicity. Thus, many scientists working on optimization of functions of hepatic cell cultures, and applicability of such cells in safety studies, are exploring the use of three-dimensional (3D) cell systems. This includes seeding matrix-like scaffolds in multiwell formats or using more advanced and organ-like setups like various bioreactors [28]. Moreover, the use of different liver cell types such as endothelial cells, Kuppfer cells, or Stellate cells in combination with hepatocytes in these models may increase assay predictability. Three-dimensional cell systems will also increase cost and complexity and decrease throughput of such model system. Therefore, it is anticipated that hPSC-derived hepatic cells for safety assessment will continue to be performed using monolayer cultures in simple formats to frontload industrial hepatotoxic testing (Fig. 1), with use of 3D cell systems as a complement to assess smaller numbers of problematic compounds and to investigate hepatotoxic mechanisms.


Ideally, one would like to evaluate safety and efficacy of new drugs in vitro in physiologically relevant cell environments. The usage of the hPSC technology is still new to the drug developing industry [29], but it holds great promise to provide a more genuine cellular context by providing access to normal human cell types unobtainable from patient sources. Furthermore, cell-based assays represent a cost-effective way to filter out unsuitable compounds (Fig. 1). In this review, we have briefly illustrated the opportunities to use hPSC-derived cardiomyocytes and hepatocytes for pharmaceutical toxicity assessment. Although there are immediate applications for the cells today, additional work must be done to standardize and validate additional assays and platforms, and at the same time increase the awareness of the technologies in the drug developing industries. In addition, further improvements of the cells and their manufacturing processes are needed. The differentiation protocols for various cell types are continuously being enhanced and optimized from the end-points of cell functionality and cost. For the time being, we emphasize that it is important to use all available sources of hPSC in the studies of the processes and mechanisms that control human cell differentiation. The advent of the induced pluripotent technology has opened parallel opportunities for investigation of embryonic stem cells, which may help to understand the pluripotent state at the molecular level. However, without the appropriate cross comparisons between results obtained from hiPSCs and hESCs, it will be difficult to dissect true biological events from artifacts occurring as a result from external manipulations of the cells.

Despite the many challenges and “unknowns” that remain, advancements made in genetic engineering of hPSC will assist in the creation of generic and customized tools for developmental biology and drug screening [30, 31]. Further improvements of the recently described possibilities to create disease-specific hPSC-derivatives, including cardiomyocytes harboring the phenotype of the Long QT syndrome [32, 33], will make it possible to investigate specific types and mechanisms of toxicity in clinically relevant in vitro systems. In addition, by combining several hPSC-derived cell types into the same assay, simulations of systemic in vivo drug effects may be considered. For example, combining metabolically active hPSC-derived hepatocytes with other cell types, such as hPSC-derived cardiomyocytes, in a micro fluidic bioreactor system that allows the metabolites from the hepatocytes exposed to a drug molecule to get in direct contact with the other cell types could be one strategy to create more advanced testing systems.

Further advancements including the development of 3D tissue models based on hPSC-derived functional cells are in progress [28]. The 3D models are obviously more complex when compared with simply culturing cells on a surface in 2D. Practical issues such as drug penetration, diffusion, and analysis read-outs need to be addressed. As biological matrix technology improves, it is anticipated that we will see much progress in this area in the near future.

Label-free measurements in which the entire assay is performed in the absence of a label are being investigated as well. These methods are thought to be more physiologically relevant and the actual biology may be better mimicked [34]. Going forward, it is also likely that these label-free screening technologies will be combined with hPSC-derived cell types for toxicity testing and there are now plate-based label-free tools that allow higher throughput analysis with increased sensitivity. Label-free measurements are typically continuous and noninvasive, so that the cells may be kept in a normal physiological state during the entire assay [34]. The assays can interrogate diverse aspects of cellular biology including receptor-mediated signaling pathway activation, proliferation, adhesion, differentiation, ion-channel activity, and transporter activity.

Other opportunities take advantage of imaging screens using automated microscopes and image analysis, commonly referred to as “high content screening,” which are now established in the pharmaceutical industries. The possibilities to perform phenotypic multiparametric cellular assays in an automated and quantitative fashion using hPSC-derived functional cells provide platforms for conducting physiologically relevant toxicity testing with exquisite time and spatial resolution. High content screening campaigns have already been conducted using hPSCs by several investigators and additional progress is expected as well [35].

The current expectations of what hPSCs can bring to the table are tremendous, but it should be remembered that the hPSC technology has only been around for about a decade and induced pluripotent technology is even more recent than that. Substantial progress has been made in a short time. Undoubtedly, the coming years will be exciting times for everyone involved in basic research and development in the hPSC field.


This work was supported by Cellartis AB, the EU-funded projects “SCR&Tox” (grant number: 266753), and “CarcinoGENOMICS” (contract number: LSHB-CT-2006-037712).


P.S. and P.B. are employed by Cellartis AB (