The New Approach Methodologies (NAMs)

Introduction: A Field at an Inflection Point

Drug development today sits at a paradox that has become increasingly difficult to ignore. Scientific capability has advanced at an extraordinary pace, yet the success rate of translating promising compounds into approved therapies remains strikingly low. More than 90% of drug candidates that appear effective in preclinical studies fail once they reach human trials, a statistic that has barely shifted over decades.

This persistent failure is not due to a lack of innovation, nor a shortage of biological insight. Instead, it reflects something more structural: the field has long relied on experimental systems that approximate human biology without truly replicating it. Animal models, which have anchored preclinical research for over a century, capture fragments of human physiology,but not its full complexity.

The 2026 Science review by Wu and colleagues arrives at a moment when this long-standing assumption is being actively questioned. Drawing from converging advances in human-derived biology, engineered systems, and computational modeling, the authors argue that new approach methodologies (NAMs) are not simply incremental improvements. They represent a shift in how drug development itself is conceptualized. This essay explores that claim, examining the biological, technological, and regulatory forces driving this transition, while also considering the challenges that remain unresolved.

The Translational Gap: Why Animal Models Fail

Understanding the importance of NAMs begins with understanding the translational gap, the disconnect between preclinical findings and clinical reality.

For decades, animal models, particularly rodents, have served as the foundation of drug discovery. Their use has been justified by assumed physiological similarities to humans. But as therapeutic strategies have evolved, that assumption has become increasingly fragile.

Modern drug modalities, including oligonucleotide therapies, antibody-drug conjugates, and targeted protein degraders, often act on mechanisms that simply do not exist in standard laboratory animals. A molecule designed to degrade a human-specific protein cannot be meaningfully evaluated in a mouse that lacks that target. In such cases, the issue is not reduced predictability, it is the absence of relevance altogether.

Even when targets are conserved, systemic differences complicate interpretation. Drug metabolism in the liver varies significantly between species, particularly in cytochrome P450 activity. Cardiac electrophysiology differs in ways that undermine arrhythmia prediction. Immune responses are shaped by distinct cytokine networks and receptor distributions. Each divergence introduces uncertainty, increasing the likelihood of both false positives, drugs that appear effective but fail clinically, and false negatives, where potentially useful therapies are abandoned prematurely.

The consequences extend beyond scientific inefficiency. Late-stage clinical failures drive development timelines into the 10–15 year range and push costs beyond a billion dollars per approved drug. NAMs are emerging not as a refinement of this system, but as an attempt to address its underlying limitations.

Human-Derived Cellular Systems: The Biological Foundation

At the core of NAMs are human cellular systems. These platforms bring drug testing closer to the biological context that ultimately matters: human tissue.

Wu and colleagues categorize these systems into three main types, primary cells, immortalized cell lines, and stem cell-derived platforms, each offering distinct advantages and constraints.

Primary and Immortalized Cells

Primary human cells retain the defining features of mature tissues. Hepatocytes preserve metabolic activity, cardiomyocytes maintain electrophysiological properties, and immune cells respond with physiologically relevant signaling. Their strength lies in fidelity: they reflect real human biology with minimal abstraction. This is why primary hepatocytes remain central to evaluating drug metabolism, particularly cytochrome P450 interactions.

Immortalized cell lines occupy a different niche. Their scalability and reproducibility make them indispensable for early-stage screening. HEK293 cells are widely used for receptor signaling studies, while HepG2 and HepaRG cells model hepatic metabolism. Tumor-derived lines such as A549 and MCF-7 provide accessible cancer models.

However, this convenience comes with trade-offs. Immortalization alters gene regulation, cellular architecture, and responsiveness. These models are useful, but only when their limitations are explicitly acknowledged and managed.

Induced Pluripotent Stem Cells: A Transformation in Disease Modeling

The emergence of induced pluripotent stem cell (iPSC) technology has fundamentally expanded what is possible in human modeling. By reprogramming adult cells into a pluripotent state, researchers can generate virtually any cell type while preserving the donor’s genetic background.

This capability allows disease modeling to move beyond generalized systems into patient-specific biology. iPSC-derived neurons, cardiomyocytes, and endothelial cells can recapitulate disease phenotypes within defined genetic contexts. Initially applied to monogenic disorders, this approach now extends to neurodegeneration, cardiovascular disease, viral susceptibility, and fibrosis.

Its impact on therapeutic discovery is already visible. Ropinirole, identified through iPSC-derived motor neuron screening, has emerged as a candidate therapy for amyotrophic lateral sclerosis. Similarly, patient-specific endothelial models revealed a role for lovastatin in LMNA-associated vasculopathy, with subsequent clinical observations supporting improved vascular function.

Yet iPSCs are not universally superior. Reprogramming resets epigenetic memory, which can erase features tied to aging or environmental exposure. In diseases shaped by chronic inflammation or long-term tissue remodeling, primary cells may better preserve relevant phenotypes. Airway epithelial cells from COPD patients and aged dermal fibroblasts, for instance, retain disease-specific characteristics that iPSCs may not.

The implication is clear: these systems are not interchangeable. Effective use requires selecting the platform that best reflects the biology of interest.

Microphysiological Systems: Reconstructing Organ Complexity

While cellular systems capture molecular detail, they cannot fully replicate the structural and mechanical context of living tissues. Microphysiological systems (MPSs) address this gap by introducing architecture, dynamics, and multicellular interactions.

Organoids: Self-Organization as a Design Principle

Organoids represent a biologically driven approach to tissue modeling. Derived from stem cells, they self-organize into three-dimensional structures that mimic key aspects of organ function.

Their diversity is striking. Intestinal organoids recreate epithelial complexity from single stem cells. iPSC-derived systems have extended this to tissues such as brain, retina, liver, and heart. These models are not engineered piece by piece,they emerge from intrinsic developmental programs.

One of the most impactful applications has been in oncology. Patient-derived tumor organoids (PDTOs), generated directly from clinical samples, preserve the genetic and structural features of original tumors. Their predictive value is increasingly evident. In colorectal cancer, organoid responses have mirrored patient responses to chemotherapy. Liver cancer organoid biobanks have identified resistance mechanisms, including c-Jun-mediated resistance to lenvatinib.

Organoids are also contributing to drug discovery itself. The development of the bispecific antibody MCLA-158, targeting EGFR and LGR5, was supported by functional validation in colon cancer organoids, demonstrating their utility beyond prediction.

Organs-on-Chips: Engineering the Microenvironment

Where organoids rely on biological self-assembly, organs-on-chips introduce controlled physical environments. Microfluidic systems simulate flow, mechanical stress, and molecular gradients, conditions absent in static cultures.

The lung-on-a-chip exemplifies this approach. By applying cyclic stretch and fluid flow across epithelial and endothelial layers, it recreates aspects of breathing physiology. Similar platforms now exist for heart, kidney, gut, and the blood-brain barrier.

These systems offer a critical advantage: dynamic realism. A patient-specific esophageal cancer chip model, incorporating tumor organoids and stromal cells, predicted chemotherapy responses more accurately than static cultures, within clinically relevant timeframes. This suggests that recreating physiological conditions is as important as using human cells.

Complexity continues to expand. Incorporating immune cells allows evaluation of immunotherapies, distinguishing targeted cytotoxicity from nonspecific activation. Microbial co-culture systems model host–microbiome interactions, as shown in cervix and vaginal chips where microbial composition directly influenced epithelial health.

Hybrid and Multiorgan Systems: Approaching Systemic Physiology

The next step is integration. Organoid-on-chip systems combine biological realism with engineered control, while multiorgan platforms connect multiple tissues into shared circuits.

An iPSC-derived multitissue system linking heart, liver, bone, and skin has reproduced pharmacokinetic responses to doxorubicin, identifying early cardiotoxic signals. More complex systems connect up to six tissues, enabling the study of metabolism, toxicity, and interorgan communication.

These platforms begin to approximate whole-body physiology, addressing a fundamental limitation of isolated models. Drugs act across systems, not within a single organ. Capturing that interplay is essential for predictive accuracy.

Artificial Intelligence: The Computational Engine

If human-derived systems provide biological relevance, artificial intelligence provides interpretive power. Together, they form a feedback loop between experimentation and prediction.

Generative AI for Molecular Design

Deep learning has transformed molecular design. Models trained on large chemical datasets can generate novel compounds optimized for multiple properties simultaneously.

The development of a TNIK inhibitor for idiopathic pulmonary fibrosis, identified and advanced to clinical testing in under two years, illustrates the acceleration possible with AI-guided design.

This approach is expanding into more complex modalities, including PROTACs and peptides, where design spaces are too large for traditional methods.

Active Learning and Closed-Loop Experimentation

Active learning introduces a dynamic element. Instead of testing compounds randomly, models select experiments that most improve predictive accuracy.

Platforms like DrugReflector have demonstrated dramatic improvements in hit rates, leading to candidates such as CLY-124 for sickle cell disease. Similarly, autonomous systems like LUMI-lab integrate AI with robotic synthesis to explore vast design spaces, identifying improved delivery systems for gene editing.

Multimodal Learning and Agentic AI

At the frontier, multimodal models integrate diverse data types, imaging, genomics, clinical data, into unified representations. Systems like BiomedCLIP and TITAN exemplify this convergence.

Even more striking is the emergence of agentic AI. Systems capable of orchestrating entire research workflows, designing molecules, guiding experiments, and refining hypotheses, are beginning to resemble autonomous research environments. The concept of a continuously learning laboratory is no longer theoretical.

An Integrated Roadmap: Four Steps to Human-Centric Drug Development

Wu and colleagues outline a four-step framework for integrating NAMs:

  • Lab-in-a-loop: Continuous feedback between experiments and AI-driven models.
  • Human organs-on-chips: System-level evaluation of pharmacokinetics and safety.
  • Clinical trial-in-a-dish: Population-scale testing using patient-derived models.
  • Digital-experimental twins: Integration of biological data with computational simulations.

Together, these steps form a coherent pathway toward human-centered drug development.

Selecting Bioanalytical Assays for FDA Approved Clinical Trials and Commercialization

Recent White paper
Selecting Bioanalytical Assays for FDA Approved Clinical Trials and Commercialization


Read the full White paper →

Regulatory Evolution and Ethical Governance

Technological readiness is advancing faster than regulatory adaptation. While legislation such as the FDA Modernization Acts 2.0 and 3.0 signals progress, full integration of NAMs requires standardized validation frameworks and global alignment.

Ethical considerations are also evolving. The use of patient-derived materials, increasingly complex in vitro systems, and algorithm-driven decision-making introduces new questions. Addressing these will require ongoing, adaptive governance.

Equally important is education. Transitioning from animal-based paradigms to NAM-driven systems demands new skill sets, combining biology, engineering, and computational thinking.

Critical Assessment

The promise of NAMs is compelling, but challenges remain.

Prospective validation is still limited. Many successes are retrospective, demonstrating what could have been predicted rather than guiding real-time decisions.

Reproducibility remains an issue. Biological variability complicates standardization across laboratories.

Integration is technically and organizationally complex, requiring coordination across disciplines.

These challenges are significant, but they are solvable. Importantly, they are not rooted in fundamental scientific limitations.

Conclusion: A Paradigm in Transition

Drug development is undergoing a shift that is deeper than technological innovation. It is a redefinition of how evidence is generated and interpreted.

NAMs move the field toward systems that reflect human biology more directly, reducing reliance on approximations and improving predictive accuracy.

If regulatory, ethical, and educational systems evolve alongside these technologies, the result could be a more efficient, transparent, and human-relevant drug development paradigm.

The long-standing gap between scientific discovery and clinical success has persisted for too long. NAMs offer a path, not guaranteed, but credible, toward closing it.

Reference

  1. Wu et al., Reimagining human -centric drug development with new approach methodologies. Science, 2026, 392: 371-378 DOI: 10.1126/science.aeb0045The New Approach Methodologies (NAMs).

Image credit: Portions of the figure in this article were generated using ChatGPT (OpenAI) or Google Gemini.

Disclaimer: This blog post is intended solely for educational and scientific informational purposes. Any mention of therapeutic drug names, including FDA-approved medications, is for the purpose of accurate reporting and discussion of biomedical research and does not constitute medical advice, endorsement, or promotion. Readers should not interpret the content as a recommendation for any specific treatment. Always consult a qualified healthcare professional for medical advice or treatment decisions.

About Marin Biologic Laboratories

Our Recent Publication/Meeting Presentation on Gene Therapy

1. Development of VNX-101, an Adeno-Associated Virus with Less Immunogenicity and Efficient Long-Term Expression of a CD19 T-Cell Engager. Molecular Therapy Methods & Clinical Development, published online July 24, 2025.

3. Cell-Based Potency Assay for Anti-CD3-Anti-CD19 Diabody. Journal of Immunological Methods. 2025. 545-114004.

3. Development of a Pharmacokinetic (PK) Mouse Serum GLP ELISA for an Anti–CD19–AntiCD3 Diabody
bioRxiv 2025.03.19.644217; doi: https://doi.org/10.1101/2025.03.19.644217.

4. American Society of Hematology (ASH) Annual Meeting 2024.
Abstract link: Using Gene Therapy to Solve Challenges with CAR-T Cell Immunotherapy: Lead Selection and Preclinical Development of an Adeno-Associated Virus with Reduced Immunogenicity Exhibiting Efficient and Long-Term Expression of an Anti-CD19 T-Cell Engager.

Comprehensive Assay Solutions for In Vitro and Cell Based Potency Assays and Pharmacokinetics Studies- Our Expertise

With 30 years of expertise in cell culture, cell-based assays, and preclinical/clinical PK/PD analysis, we specialize in offering assay services essential for a wide variety of therapeutic drug development programs, preclinical studies, IND/BLA applications, and commercialization. Our comprehensive services include both preclinical non-GLP and GLP assays, as well as non-GMP and GMP assays, providing critical support throughout the entire development pipeline.

Watch the following video and explore our latest presentation on the development and validation of potency and pharmacokinetic (PK) assays for AAV vectors, highlighting innovative methodologies and industry-leading expertise.

 

 

Download the full presentation: Development of Custom Cell Based and In vitro Potency and Pharmacokinetics (PK) Assays for AAV vectors- Marin biologic Laboratories

 

Development of Cell-Based Potency Assays: Case Studies and Blogs from Marin Biologic Laboratories (MarinBio)

 

Drug Discovery & Development Assays Offered by Marin Biologic Laboratories (MarinBio)