A compound might look safe in a dish but cause arrhythmias in a real patient.
For generations, the search for safer heart medicines has been constrained by tools that only approximate the organ they seek to understand — rodent hearts that beat to a different rhythm, and flat cell cultures that cannot hold the architecture of life. Researchers in Shanghai and their collaborators have now mapped how three-dimensional cardiac constructs, grown from human stem cells, are beginning to bridge that ancient gap between the laboratory bench and the living chest. These engineered tissues contract, conduct electricity, and respond to drugs in ways that more faithfully mirror the human heart — though the path from promising model to regulatory acceptance still winds through unsolved problems of maturation, vascularization, and standardization.
- Decades of cardiovascular drug failures trace back to a fundamental mismatch: the models used to test heart medicines — rodents and flat cell cultures — cannot fully replicate what happens inside a human heart.
- Three-dimensional cardiac constructs grown from human stem cells now contract with measurable force, fire coordinated electrical signals, and carry patients' own disease mutations — making them far more faithful stand-ins than anything that came before.
- Patient-derived tissues open a new frontier in personalized medicine, allowing researchers to test drugs directly on tissue that carries the genetic signature of a specific inherited heart disease.
- AI and advanced biomaterials are accelerating development, but the field is still wrestling with tissues that behave like fetal hearts rather than adult ones, lack blood vessels, and produce results that vary from lab to lab.
- Widespread adoption in drug discovery pipelines remains contingent on solving maturation and reproducibility challenges, establishing standardized protocols, and convincing regulators to recognize these models as credible evidence.
For decades, pharmaceutical researchers have tested heart drugs using two imperfect proxies: rodent models whose cardiovascular biology diverges meaningfully from our own, and flat cell cultures that strip away the architecture, electrical coordination, and mechanical forces that define a living heart. The gap between these tools and the actual human organ has long been a quiet crisis in drug discovery — compounds that look safe in a petri dish or a mouse can fail catastrophically, or cause unexpected harm, when they reach a patient.
A comprehensive review published in the journal Research by researchers at Shanghai University and their collaborators documents how three-dimensional cardiac constructs — engineered tissues grown from human stem cells — are beginning to close that gap. These models take several forms: scaffold-based tissues, suspension cultures that self-assemble, microfluidic heart-on-chip devices that mimic blood flow, and scaffold-free microtissues suited for high-throughput drug screening. Cardiac organoids, which develop with minimal external guidance, are especially valuable for studying inherited heart diseases and fetal cardiac development.
Perhaps the most consequential capability is disease modeling. Researchers can reprogram cells from patients with inherited cardiomyopathies into pluripotent stem cells, grow 3D cardiac tissue carrying the patient's own mutations, and test drug candidates directly on that tissue. For non-genetic conditions, disease states can be induced through metabolic stress, inflammation, or oxygen deprivation — creating models of heart attack or acquired disorders.
Measuring what these tissues do requires sophisticated tools: multielectrode arrays, calcium imaging, optical mapping, and metabolic analysis. But many of these techniques were designed for flat cultures, and the field still lacks standardized protocols for 3D systems — a gap that complicates cross-laboratory comparison. Biomaterials science and 3D bioprinting are improving how cells are supported and arranged, while artificial intelligence is being applied to image analysis and model design, with the potential to filter out weak drug candidates before costly lab work begins.
Significant challenges persist. Stem cell-derived heart cells often behave like fetal rather than adult tissue, affecting how they handle calcium, conduct electricity, and respond to drugs. Most constructs lack vasculature, limiting their size and longevity. Reproducibility remains inconsistent across cell batches and laboratory conditions. The researchers are clear that 3D models should complement — not replace — animal studies and traditional cultures. Realizing their full potential will require solving maturation and vascularization problems, standardizing protocols, reducing costs, and establishing regulatory frameworks that give pharmaceutical companies a clear path forward.
For decades, the pharmaceutical industry has relied on a familiar pair of tools to test whether a new heart drug is safe and effective: mice and rats in laboratories, and petri dishes filled with human heart cells grown flat against plastic. Both have serious limits. Rodents are rodents—their hearts beat differently, metabolize drugs differently, respond to compounds in ways that don't always translate to human patients. Flat cell cultures, meanwhile, are just that: flat. They cannot capture the three-dimensional architecture of living cardiac tissue, the electrical signals that coordinate a heartbeat, the mechanical forces that make the heart contract, or the intricate conversations between different cell types that keep the organ alive.
This gap between what we can test in the lab and what actually happens in a patient's chest has long been a bottleneck in cardiovascular drug discovery. A compound might look promising in a petri dish or pass safety tests in mice, only to fail or cause unexpected harm when given to humans. Researchers at Shanghai University and their collaborators have now published a comprehensive review in the journal Research documenting how three-dimensional cardiac constructs—engineered tissues grown from human stem cells—are beginning to close that gap.
These 3D models are built using several different approaches. Some use scaffolds, physical frameworks that guide cells into organized structures. Others rely on suspension culture, where cells self-assemble in liquid. Still others employ microfluidic devices—tiny channels that mimic blood flow—or allow cells to organize themselves without any external support. What matters is the result: tissues that behave far more like actual human heart tissue than anything a flat culture can achieve. They contract with force. They fire electrical signals in coordinated patterns. They metabolize nutrients and respond to drugs in ways that reflect what happens inside a living heart.
The review identifies four major types of 3D cardiac models, each with distinct strengths. Engineered heart tissues excel at measuring how forcefully the tissue contracts and how it responds to mechanical stress. Heart-on-chip systems—miniaturized devices that recreate the heart's environment—allow researchers to apply controlled flow and dynamic stimulation, mimicking the pulsing action of blood vessels. Scaffold-free cardiac microtissues can be produced in large numbers, making them useful for screening many drug candidates at once. Cardiac organoids, which develop from stem cells with minimal guidance, are particularly valuable for studying how the heart forms during fetal development and for modeling inherited genetic diseases.
That last capability opens a powerful door. Researchers can now take cells from patients with inherited heart conditions—dilated cardiomyopathy, hypertrophic cardiomyopathy, arrhythmogenic cardiomyopathy, and others—reprogram them into pluripotent stem cells, and grow 3D cardiac tissue that carries the patient's own disease-causing mutations. They can then test drugs on that tissue to see which compounds might help that specific patient. For conditions that aren't genetic, researchers can artificially induce disease states by exposing the tissue to metabolic stress, inflammatory signals, toxins, or low-oxygen conditions, creating models of heart attack, inflammation, or other acquired disorders.
But building a better model is only half the battle. The review emphasizes that these tissues must also produce reliable, quantifiable measurements. Researchers need to know whether a drug affects the heart's electrical rhythm, its contractile force, its energy metabolism, or its vulnerability to toxicity. Technologies like multielectrode arrays, optical mapping, calcium imaging, and metabolic flux analysis can provide these readouts. Yet many of these measurement techniques were originally designed for flat cultures, and the field still lacks standardized protocols tailored to intact 3D tissues—a gap that slows adoption and makes it hard to compare results across laboratories.
Two emerging forces are accelerating progress. Biomaterials science is producing conductive polymers and hydrogels that better support cardiac function, while 3D bioprinting allows researchers to position multiple cell types in precise spatial arrangements. Artificial intelligence is being harnessed to analyze images of the tissues, identify different cell types, and even help design better models. Because growing and testing 3D cardiac constructs remains expensive and time-consuming, AI workflows that can filter out weak drug candidates before they reach the lab could dramatically speed up discovery and lower costs.
Yet significant obstacles remain. Stem cell-derived heart cells often behave like fetal or newborn heart tissue rather than mature adult myocardium—a difference that affects how they handle calcium, use energy, conduct electricity, and respond to drugs. Most 3D constructs lack blood vessels, which limits how large they can grow and how long they can survive in culture. Reproducibility is inconsistent; results vary depending on which cells are used, how efficiently they differentiate, which batch they come from, and how the assay is performed. The researchers argue that 3D cardiac models should complement, not replace, animal studies and traditional cell cultures. Widespread adoption in drug discovery will require solving these maturation and vascularization challenges, standardizing manufacturing and measurement protocols, reducing costs, and establishing clear regulatory pathways that pharmaceutical companies can follow with confidence.
Notable Quotes
3D cardiac constructs should complement, rather than replace, existing animal and in vitro models— Researchers at Shanghai University, published in Research
The Hearth Conversation Another angle on the story
Why does it matter that we move beyond flat cell cultures? Aren't they already pretty good at telling us if a drug is toxic?
Flat cultures are useful, but they're missing almost everything that makes a heart a heart. No three-dimensional structure, no electrical signals coordinating between cells, no mechanical forces. A drug might look safe in a dish but cause arrhythmias or heart failure in a real patient.
And rodent models—why can't we just stick with those?
Rodents are valuable, but their hearts are fundamentally different from ours. They metabolize drugs differently, their electrical systems work differently, their disease patterns don't always match human disease. A compound safe in a mouse might harm a human heart.
So these 3D constructs solve both problems at once?
They get much closer. They're made from human cells, they have actual three-dimensional architecture, they conduct electricity, they contract. But they're still not perfect—they're often more like fetal heart tissue than adult tissue, and they don't have blood vessels yet.
What's the practical advantage for a drug company?
They can test more compounds faster and more accurately before investing in expensive human trials. They can also model a patient's specific genetic disease and test which drugs might work for that person. That's personalized medicine becoming real.
What's holding them back from being used everywhere right now?
Cost, time, reproducibility, and lack of standards. Different labs get different results because there's no agreed-upon way to build or measure these tissues. Regulators also haven't yet established clear pathways for using them in drug approval. Until those problems are solved, most companies will keep using the old methods.