Validating and Using Cardiac NAMs for Toxicity Screening and Drug Development

Abstract

Technological advances and the desire to reduce dependence on animal models have brought human-relevant models to the forefront of drug development. This paradigm shift is leveraging the advances in in vitro systems and new approach methodologies (NAMs), which was the focus of a workshop convened by the Health and Environmental Sciences Institute (HESI) in May 2024. Highlights included discussions on predicting cardiac failure modes and the utility of human-induced pluripotent stem cell–derived cardiomyocytes (hiPSC-CMs), microfluidic systems like BioFlux™, and engineered heart tissues in enhancing early-stage drug safety assessments. Regulatory perspectives underscored the challenges and potential for integrating NAMs into submissions, advocating for standardized reporting and validation protocols. Case studies where NAMs offered superior predictivity compared to traditional methods are emerging and offer insights into a roadmap forward. However, there remains a need for collaboration among academia, industry, and regulatory bodies to ensure robust validation and adoption. These efforts aim to refine cardiovascular drug discovery, reduce attrition rates, and accelerate the transition toward more ethical and efficient preclinical testing paradigms.

Introduction

Traditional approaches to assess cardiac safety liabilities in drug development often rely on animal models that have challenges, namely, scalability, predictivity to human outcomes, and lack of mechanistic insight. A new paradigm could capitalize on the known mechanisms of cardiac liabilities as well as novel in vitro and in silico technologies to shift screening of cardiac safety liabilities earlier in the drug development process allowing for earlier derisking, lead candidate optimization, and more targeted animal assessments. ¹ In order to help support this paradigm shift, Health and Environmental Sciences Institute (HESI) sought and was awarded a multi-year U01 grant from the US Food and Drug Administration (FDA) on Validating Human Stem Cell Cardiomyocyte Technology for Better Predictive Assessment of Drug-Induced Cardiac Toxicity.

HESI is a global nonprofit organization that aims to address global health and environmental challenges through collaborative science. This is accomplished largely through scientific committees of public–private experts tasked with collaborating on translating new science into applied solutions. One of these committees, the Cardiac Safety Committee, is dedicated to improving public health by reducing unanticipated cardiovascular (CV)-related adverse effects from drugs or chemicals. The Cardiac Safety Committee recognized the need for a modernized paradigm shift to focus on specific cardiotoxic mechanisms that contribute to drug attrition.

A growing number of biological frameworks have been proposed to rationalize and guide the development of a novel assessment paradigm, each addressing key aspects of how toxicity arises. Specifically, toxicity is the outcome of a few fundamental features of a xenobiotic: its bioavailability to the host, its biological interaction with the host, and a response that surpasses the host’s ability to adapt to that exposure (i.e., adversity). ² Experts in the Cardiac Safety Committee previously described the seven cardiac failure modes: vasoactivity, contractility, rhythmicity, myocardial injury, endothelial injury, vascular injury, and valvulopathy. These failure modes represent the finite number of ways that an organ system adversely responds to a xenobiotic. ¹ The cardiac failure modes helped frame the HESI U01 award and subsequent research. Through this award, HESI sought to test new approach methodologies (NAMs) that could identify one or more of the cardiac failure modes through a series of pilot studies and subsequent validation studies.

In the final year of the U01 grant and in support of advancing novel cardiac assays, HESI convened a 1-day workshop in May 2024 at the US FDA White Oak campus in Silver Spring, MD, USA. The workshop focused on the results of the U01 grant validation studies and use of cardiac NAMs for toxicity screening and drug development. While NAMs have been described in a number of ways, for the purpose of the workshop the Interagency Coordinating Committee on the Validation of Alternative Methods’ previously adopted description was used, which describes the term NAMs as “any technology, methodology, approach, or combination thereof that can be used to provide information on chemical hazard and risk assessment and supports replacement, reduction, or refinement of animal use (3Rs).” ³ The workshop brought together a diverse group of experts to address critical gaps in cardiac toxicity testing within drug development, focusing on leveraging NAMs. The discussions emphasized improving predictive models and streamlining regulatory processes to enhance safety and efficacy evaluations.

Dr. Brian Berridge, B2 Pathology Solutions, presented his perspectives on the need to modernize drug screening and toxicity testing. As one of the authors of the cardiac failure modes, ¹ Dr. Berridge emphasized the importance of a paradigm shift. He advocated that our current efforts to re-invent preclinical drug safety assessment are largely driven by advancements in new technologies rather than the intentional design of solutions tailored to specific scientific questions. The specific questions being addressed and the decisions being informed have remained fundamental and fairly consistent throughout time in drug safety assessments. Nevertheless, these assessments have evolved in response to an evolved understanding of adverse effects, with newly recognized effects prompting additions to existing evaluation protocols. Importantly, these cumulative experiences in evaluating drug safety and characterizing CV responses to drug-induced injury using current approaches offer valuable insights for developing a more efficient, mechanistically informative approach to CV drug safety assessment.

Adverse outcomes pathways (AOPs) describe the mechanistic steps leading to those failure modes, ⁴ and key characteristics are bioactivities of xenobiotics that have most often been associated with toxic responses in the host. ⁵ Integrating the mechanistic and biological frameworks of the failure modes and AOPs establishes the foundation for developing a novel assessment paradigm, encompassing the definition of biological scope, assessment conditions, informative mechanistic endpoints, and decision thresholds. By leveraging advances in in vivo–relevant cell-based modeling systems within an evidence-based framework—grounded in decades of safety assessment experience—this approach could be applied earlier in the development process, providing improved guidance for risk mitigation strategies.

Emerging Technologies

There are numerous new and emerging technology platforms available to employ for CV toxicity assessments. Key considerations for using such models in regulatory applications include defining a clear context of use (COU) for new drug applications for in vitro models, which should be simple yet sufficiently complex to address the targeted questions. Additionally, standardization of cell culture conditions and incorporation of appropriate quality controls, both positive and negative, are critical to ensure model performance, enhance reproducibility, and improve predictability in assessments.

With this in mind, research is needed to provide data to understand the reliability, validity, and limitations of new assays. HESI, FDA, and many other stakeholders are working to provide this needed data. At the workshop, several NAMs were shared along with preliminary data as exemplars in this space.

HESI Grant-Funded Studies

Drs. Gina Smith and Daniel Bramham of Labcorp received an award for the development of an in vitro microfluidic model of the human CV system. The work aimed to address the lack of regulatory in vitro assays for predicting atherosclerotic risk following drug or chemical exposure and to address the cardiac failure mode of vascular injury. The model consisted of human aortic endothelial cells, which mimic the vasculature, and THP-1 monocytes, which simulated the immune response. The model is built on the BioFlux™ system; a microphysiological system capable of applying precise shear flow to the model vessels, and providing a physiologically relevant environment in which to assess drug or chemical-linked effects on the endothelium.

Labcorp initiated the study by first defining appropriate dose levels, using qualitative and quantitative endpoints to assess toxicity in the microfluidic channels following chemical exposure. For the assessment of atherosclerotic risk of compounds, the team developed and validated assay endpoints for monocyte adhesion, a critical factor in atherosclerosis, as well as cytokine release. The results demonstrated the model’s capacity to predict both pro- and anti-inflammatory responses following exposure to pharmaceuticals, food additives, and chemical compounds.

The prediction of the in vitro model correlated well with literature expectations, which highlighted its potential for use in pharmaceutical and chemical screening, target identification, and lead compound optimization.

The second study, awarded to Dr Joseph C. Wu at Stanford University, highlighted unique benefits of induced pluripotent stem cell (iPSC) models in CV research. These models offer genetic similarity to patients and high expansion potential, making them well-suited for applications such as drug testing, disease modeling, cell therapy, and 3D tissue engineering. The study focused on predicting vascular liability using transcriptome analysis, a method for detecting drug-induced vascular toxicity, often missed by in vivo studies and addresses the cardiac failure mode of vascular toxicity. This research demonstrated that iPSC-derived endothelial cells can recreate vascular phenotypes in vitro, providing a reliable surrogate for primary vasculature and improving toxicity detection.

Additional lab work at Stanford, not funded by the U01, looked at using 3D engineered heart tissues (EHTs) to study tachycardia-induced cardiomyopathy (TIC). This condition, characterized by a persistently high heart rate leading to heart failure, is mostly reversible, unlike other heart failures. Traditional animal models have limitations in throughput, cost, and ethical concerns. In this study, EHTs were used to simulate chronic tachycardia, revealing reversible contractile dysfunction and hypoxia, similar to clinical findings. The sequencing and analysis of hypoxia markers identified nicotinamide adenine dinucleotide (NAD) homeostasis as a crucial factor. Restoring NAD levels was shown to accelerate tissue recovery, offering insights into therapeutic strategies for TIC. ⁶

The final study, awarded to Dr Alex Savtchenko at Nanotools Bioscience, focused on developing a platform for drug discovery, with specific applications in cardiotoxicity and an aim to address the cardiac failure mode of rhythmicity. Current in vitro methods for investigating electrophysiological cellular responses often have limitations and are conducted under conditions that fail to mimic specific cardiac events, such as arrhythmias. To address these challenges, Nanotools introduced an innovative approach using physiological cell stimulation with graphene as a mediator. The technology employs light to stimulate cardiomyocytes offering a more flexible and precise screening method compared to traditional drug discovery techniques.

Specifically, the technology harnesses the unique optoelectronic properties of graphene, including its ability to efficiently transduce light into electrical signals without generating significant heat or altering pH, thereby preserving cell viability. This capability enables precise, non-invasive modulation of cardiomyocyte activity and facilitates the study of drug effects under various physiological and pathological conditions, including arrhythmias. The system has been validated through extensive testing, ⁷ demonstrating its reliability and reproducibility. In addition, custom software tools have been developed to quantify light-induced activation patterns and complex pharmacological responses in cardiomyocytes.

Ongoing work is focused on extending the technology to 3D tissue models and validating its performance across a broader range of therapeutic compounds. We anticipate that graphene-enabled optical stimulation of cardiomyocytes will provide a powerful, next-generation platform for drug discovery, enhancing the predictive power of in vitro models for the development of more effective anti-arrhythmic therapies.

Inflammatory Response Model

Dr. Li Pang from National Center for Toxicological Research, FDA, highlighted the importance of immune response in toxicity and presented the idea of developing human-based immuno-cardiac models for regulatory purposes. As the landscape of therapeutic development advances, predicting immunotoxicity in novel modalities has become a critical challenge. Over recent decades, novel therapeutic modalities, such as protein-targeted chimerics, new-generation peptides, large peptides, nucleic acid–based therapeutics, and biologics, have seen increased approval rates, accounting for 48% of FDA approvals between 2021 and 2022. ⁸ Compared to traditional small molecules, some of these new therapeutic modalities pose a higher risk of inducing severe immune adverse reactions. To address potential immune liabilities of therapeutics that target or may affect the immune system, the updated FDA guidance on Nonclinical Evaluation of the Immunotoxic Potential of Pharmaceuticals requires preclinical studies to assess modes of action and provide evidence-based risk assessments for intended or unintended immune effects. ⁹

Although infrequent, CV toxicity associated with immunotherapy presents a significant risk of morbidity and mortality. Considering the species differences in immune and CV systems, the elevated risk of immunogenicity with novel therapeutic modalities, and the growing prominence and application of immunotherapy, there is an urgent need to develop human cell–based NAMs integrated with immunology to enhance the prediction of immune-related CV safety. ¹⁰ Incorporating immune components into novel cardiac models for drug discovery and evaluation is a key piece to understanding immunotoxicity. Dr. Pang shared both the potential and challenges of developing a co-culture using human-induced pluripotent stem cell–derived cardiomyocytes (hiPSC-CMs) and macrophages to evaluate inflammation-related myocyte damage.

Cardiac NAMs in Practice

While the traditional cardiac safety testing paradigm is rigorous, drug-induced cardiac toxicity is still a well-documented cause of drug attrition during late development stages or post-marketing approval. New methodologies aim to identify potential cardiac liabilities earlier in the drug development process, ideally in the process of lead candidate selection but also in the course of early clinical studies. The intent of expanding use of these new methodologies is to reduce the risk of late-stage failures, improve patient safety, and reduce animal use. Should these new methodologies show less propensity for “false positives” and greater translational relevance to human subjects, they may also find utility in derisking safety signals that arise from conventional animal studies, thus reducing attrition of potentially efficacious investigational drugs. However, establishing consensus on standards and practices is critical to ensure any new technologies are fit-for-purpose.

HESI has collaborated with FDA to standardize cardiac NAMs and hiPSC-CMs for broader use during the drug development process. However, there is still a need for detailed and standardized reporting in these studies to enhance their utility and credibility. Dr. Natalie Simpson from the FDA shared some statistics from Investigational New Drug (IND) submissions since 2019, ¹¹ noting that while many studies still rely on traditional in vivo assays, there is a growing interest in using hiPSC-CMs due to their predictive capabilities, especially for drugs targeting the sodium channels or affecting heart rate. She also noted that many studies lacked critical details, such as test article concentrations and criteria for rejecting plates, which are crucial for regulatory reviews. Additionally, the reasons that a sponsor opted to conduct a hiPSC-CM study were often absent or otherwise not apparent based on data from conventional studies. In other cases, the rationale was rather evident, even if not explicitly stated, and typically centered on derisking a finding from conventional studies. It is a reasonable assumption that sponsors found value in conducting the hiPSC-CM studies and providing a more explicit explanation of that value in regulatory submissions would further FDA’s understanding of useful contexts of use. To fully integrate hiPSC-CM data into traditional safety assessments and replace certain animal studies, submissions must become more comprehensive and justified, supporting a smoother transition toward modernized cardiac safety testing.

Cardiac NAM applications span the drug development process and, in some cases, are being used to support regulatory decision-making.^11,12 Dr. Xi Yang, RTI International, shared one example where cardiac NAMs played a confirmatory role in characterizing metabolites in structural cardiotoxicity. In this example, the cardiotoxic metabolites were present in one nonclinical species at a 20-fold higher metabolite-to-parent ratio compared with humans and were 10-times more cardiotoxic to iPSC-CMs, which supported escalating doses above a traditional no observed adverse effect level–based exposure limit identified in the nonclinical species of interest. This highlights that cardiac NAMs can reveal significant differences in metabolite toxicity that are crucial for guiding clinical trials.

Dr. Yasunari Kanda, National Institute of Health Sciences (NIHS), Japan, provided some perspectives on the application of NAMs in the field of drug development and CV safety assessment. One focal point is using hiPSC-CMs as a cardiac NAM for mechanism-based approaches to improve predictability and accelerate regulatory decision-making.

One detailed case study explored drug-induced heart failure using the compound BMS-986094, which targets the hepatitis C virus RNA-dependent RNA polymerase nonstructural protein 5B and was terminated due to unexpected cardiotoxicity. ¹³ Multiple participants in the Phase II clinical trial experienced serious cardiac adverse events, including reduced left ventricular ejection fraction, and in some cases, heart failure, requiring hospitalization. This case illustrates the limitations of conventional nonclinical models and prompted an investigation into whether contractility impairment could be predicted using contractility assays with human iPSC-derived cardiomyocytes as a cardiac NAM. Using image-based motion analysis, NIHS investigators found that BMS-986094 decreased contractility in hiPSC-CMs after 96 hours of exposure. In contrast, sofosbuvir, a clinically approved drug that also targets the same protein, had little effect on contractility. These results suggest that a chronic cardiotoxicity assay using hiPSC-CMs can predict drug effects on contractility. To further confirm the usefulness of motion analysis using hiPSC-CMs, researchers compared the in vitro data with FDA Adverse Event Reporting System datasets and found that motion impairment correlated with high odds of cardiotoxicity. ¹⁴ Thus, chronic cardiotoxicity assays using hiPSC-CMs could be a predictive NAM during drug development. There remains a need for continued development and standardization of NAMs to improve their applicability and accuracy in predicting human-specific outcomes.

Dr. Khuram Chaudhary, Bristol Myers Squibb, provided some perspectives as a drug developer and noted that NAMs are not new to the pharmaceutical industry but their adoption and refinement is an ongoing process. The drug discovery and development pipeline requires extensive resources and time that can often feel like an exercise in failure due to high attrition rates primarily caused by safety and efficacy issues. NAMs can play a role in optimizing this process, potentially reducing time and costs by refining early-stage development and enhancing the quality of compounds entering clinical trials. NAMs have increasingly been used by industry to complement traditional animal testing as well as fill gaps where traditional methods have failed.

Some of the more recent advancements have included use of stem cell–derived cardiomyocytes and engineered tissues for long-term toxicity assessments, as was this case with BMS-986094 described above. In silico, computational, and artificial intelligence/machine learning could also play a role in future advancements that could enhance drug design and reduce development timelines. Despite these promising advancements, there are existing limitations, such as the complexity of validating new models, predicting human doses, and understanding chronic treatment toxicities. Further, verifying drug concentrations, assessing protein binding, and the development of improved extrapolation methods are important in interpreting results that translate complex, human biology. Collaborative efforts are needed to effectively address these challenges, and to increase the acceptance of NAMs in regulatory decisions.

Perspectives on Regulatory Acceptance

Complex in vitro models hold significant promise as tools in drug development, offering various benefits such as improved compound screening, safety assessment, and insight into drug toxicity. These models align with the principles of the 3Rs (Replacement, Reduction, and Refinement) by potentially reducing or replacing animal studies. The qualification process for these models involves rigorous evaluation to ensure reliability and applicability in drug development and regulatory review, as outlined in the 21st Century Cures Act. ¹⁵

Dr. Jeffrey Siegel of the FDA highlighted the Innovative Science and Technology Approaches for New Drugs (iSTAND) pilot program as an example of the FDA’s commitment to fostering innovation in drug development by accelerating the incorporation of novel scientific approaches. ¹⁶ Qualified models can be publicly available and used in regulatory submissions without the need for additional confirmation by the FDA, providing a streamlined pathway for their adoption. Defining the COU is critical to ensure these models address specific regulatory needs and contribute to improving safety assessments in humans. Examples such as predicting drug-induced seizures or rare toxicities highlight the potential of these models to complement or replace existing methods. The qualification process involves submitting a comprehensive package demonstrating the model’s accuracy and relevance, which, once approved, allows its use in drug development programs. While challenges exist, such as ensuring consistency across different assays, the qualification program offers a structured framework for integrating complex in vitro models into regulatory decision-making, ultimately advancing drug safety assessment and reducing reliance on animal testing.

The Food and Drug Omnibus Reform Act of 2022 (FDORA), often referred to as the FDA Modernization Act 2.0, updated FDA regulations to expand nonclinical testing options. ¹⁷ The act replaced the term “preclinical tests (including tests on animals)” with “nonclinical tests,” defined as methods conducted in vitro, in silico, in chemico, or through nonhuman in vivo tests, during or before clinical trials to evaluate drug safety and effectiveness. These methods include cell-based assays, organ chips, microphysiological systems, computer modeling, bioprinting, and traditional animal tests.

Dr. Ronald Wange formerly of the FDA provided an overview of the implications of FDORA, emphasizing that the act did not eliminate a requirement for animal studies because no such requirement existed in FDA regulations before the act. Instead, FDORA formalized the Agency’s acceptance of alternative nonclinical methods. Dr. Wange emphasized that media reports that FDORA eliminated a requirement for animal studies to support drug development reflected a misinterpretation of the impact of the change in statutory language and more importantly a misunderstanding of pre-FDORA FDA practice regarding the types of preclinical tests accepted by the Agency in support of drug development. There are already some contexts of use where non-animal methods are already routinely accepted. He stressed that it is the state of the science, and whether an alternative method can adequately ensure the safety of the clinical trial participants and those prescribed the drug after approval, that determines whether an animal study is the most appropriate approach for assessing a particular toxicological endpoint, not any limitations in FDA regulations or US statute. Dr. Wange was optimistic that the language in FDORA could lead to increased investment in the development of non-animal alternative testing methods, given the explicit support for these approaches articulated in the statute.

Dr. Wange also addressed challenges related to the validation and acceptance of NAMs, particularly in the context of international collaboration and regulatory harmonization efforts. He acknowledged the need for greater efficiency in the validation process and proposed improvements, such as sharing evidentiary dossiers and establishing standardized protocols. Despite the complexities, there remains a potential for NAMs to revolutionize drug development, provided there is continued collaboration, investment, and alignment across regulatory agencies and stakeholders.

Discussion

The joint HESI/FDA workshop highlights a commitment to advancing CV safety testing methodologies with an aim to reduce drug-induced cardiotoxicity through NAMs. The innovative assays presented at this workshop, including the BioFlux™ microfluidic systems and iPSC-derived models for vascular toxicity, hold promise for utility in drug discovery and lead optimization, in addition to potentially serving as safety studies that address theoretical or observed CV hazards.

CV NAMs focused on ion channel activity are a key component of defining the pro-arrhythmic potential of investigational drugs prior to human clinical trials. These assays have successfully reduced the occurrence of drug-induced rhythmic toxicities, such as torsade de pointes. However, CV toxicity can arise from multiple failure modes unrelated to drug interactions with cardiac ion channels. ¹ Assessing the potential of a candidate drug to engage other modes of cardiac failure in a preclinical program still relies heavily on animal testing, where the apical outcome of a failure mode (e.g., cardiomegaly, thrombosis, and cardiomyopathy) may be detected.

Despite advancements in CV NAMs that could address these failure modes—such as those highlighted at the workshop—data from such studies are rarely included in regulatory submissions. For instance, as presented by Dr. Simpson, hiPSC-CM studies accounted for only about 1% of IND submissions reviewed by the FDA during the examined period. Moreover, these submissions often lacked clear rationales for conducting the studies or explanations of how the results resolved an identified or anticipated safety concern. However, we know that sponsors are using CV NAMs in pre-regulatory toxicology and even pre-candidate selection. Factors contributing to a lack of regulatory data submission may include lack of a regulatory mandate and regulatory uncertainty. Uncertainty around data standards and qualification requirements and the perceived lack of added value may present additional roadblocks. Regardless of cause, regulatory submission of NAM studies, in parallel with traditional assessments, is encouraged. Including such studies would increase familiarity with these platforms and help both regulators and sponsors determine whether results from currently available CV NAMs add value to contemporary testing methods.

For developers of novel NAMs, an initial understanding of where their methodology would likely provide the most value within the drug development process is an important consideration. An early distinction should be made between the pre-regulatory and regulatory phases of development as priorities and objectives can differ between these stages. As performance and qualification data accumulate, the role of an NAM may expand to new applications beyond its original context. Defining its intended use early helps identify the end-users of the assay, which in part determines the level of qualification or validation required. Pinpointing the specific gap the assay fills within the broader discovery and development process, and which stakeholders will rely on the data, can guide whether formal regulatory review for qualification is necessary.

As discussed in the workshop, formal regulatory review is not a prerequisite for leveraging data from an NAM in a regulatory submission, and it is not necessary for application in the pre-regulatory space. However, NAMs that undergo formal regulatory review, such as through the iSTAND program, ensure regulatory acceptability of the data for its given COU, and facilitate wider adoption of the methodology. A particularly promising area for identifying contexts of use and advancing human-based NAMs in the regulatory space lies in scenarios where species-specific differences either limit or preclude traditional safety testing in animals. Identifying such contexts of use could accelerate the integration of NAMs into regulatory frameworks, supporting safer and more efficient drug development.

Conclusions

This workshop underscored the importance of strategic and collaborative efforts to effectively integrate NAMs into use for the prediction of drug-induced cardiotoxicity. Participants agreed NAMs offer promising insight; however, they are complex and will require additional work and validation to meet regulatory standards. Workshop discussions highlighted that collaboration among academic, regulatory, and industry stakeholders is essential to overcome scientific and practical challenges, ensuring that NAMs provide accurate and reliable data for decision-making in drug safety.

Speakers emphasized the need for ongoing work and partnerships to better understand and characterize the evolving technologies and advance the field. With a commitment to this collaborative approach, the field can continue to make strides in using NAMs to predict cardiotoxicity, resulting in safer medicines and reducing our reliance on animals.