When Jeroen de Groot joined Ncardia in 2024, he stepped into a moment that many in drug discovery have been anticipating but few have navigated well. The industry has spent decades refining animal models, adjusting protocols, and searching for signals that would bring preclinical data closer to clinical reality. The limits of that approach have become increasingly difficult to ignore. Human biology, it turns out, is the only reliable proxy for human outcomes.
Jeroen saw this shift coming. His career began at TNO in the Netherlands and spanned two decades of contract research, including a period at Galapagos and later serving as Chief Operating Officer for early discovery at Charles River Laboratories. He managed teams across the United States, the United Kingdom, Belgium, and the Netherlands. The work was ambitious and the scale impressive, but as the organizations grew, decision cycles inherently stretched. Innovation slowed.
Stepping into Ncardia offered something different. The company builds its platform on induced pluripotent stem cell technology, using controlled bioreactor differentiation to create consistent, human-relevant cell models. To de Groot, the appeal was simple. The field is moving toward human prediction. He wanted to be where that transition is being built in real time.
It is a shift that reorders priorities in preclinical research. Cell phenotype and functional markers matter more. Batch variation must be managed. Reproducibility can no longer be treated as an aspiration. Decisions depend on it.
Many laboratories still rely on donor-derived primary cells or immortalised lines. Both bring complications. Primary cells vary from donor to donor, and immortalised lines often drift from physiological reality. iPSC-derived models offer a more stable path. When grown in 3D bioreactors that control oxygen tension, pH, and differentiation conditions, they can be manufactured in large quantities, with low batch-to-batch variation and high consistency.
These principles move from theory to practice when clients bring complex questions. Neurizon, an Australian company developing a central nervous system therapy approached Ncardia seeking clarity on the mechanism of action. The therapy was already in clinical development. The team needed a model that reflected human biology rather than approximating it. By running multiple iPSC-based CNS assays, Ncardia uncovered patterns that helped explain the effects observed in patients. The work informed the client’s scientific direction and later became the subject of a conference poster and a public webinar.
Jeroen is careful to emphasize that the technology is powerful but not automatic. Not every iPSC line differentiates cleanly. Historically, 90-95% work on the first attempt or after (minor) optimization of the process, and 5-10% fail to produce the target cell type. This is where the company’s approach to partnership becomes central. Senior scientists join the earliest discussions. Feasibility is assessed before a project begins. Pilots are built in. Go and no-go criteria are established upfront. If a client line fails to meet standards, the team adjusts or reruns before considering a full programme.
The same discipline applies to validation. Each batch is assessed with phenotypic and functional tests that match the cell’s intended purpose. Cardiomyocytes must express appropriate markers, show consistent beat rates, and respond correctly to reference compounds. The process is grounded in literature, shaped by small-scale optimisation, and supported by clear checkpoints that prevent programmes from drifting off course.
The impact is not limited to mechanistic insight. Some clients now use iPSC-derived assays to streamline early decision-making. One reported that Ncardia’s models predicted their animal outcomes with about 80% accuracy, providing a faster route to triage compounds and reducing reliance on animal testing. For de Groot, this is evidence of a broader change sweeping through the industry. Pharmaceutical companies have continued to source iPSC-based models because they influence real choices. If the data were not advancing decisions, demand would not persist.
Looking ahead, de Groot sees the most immediate opportunity in areas where animal models have long been unsatisfactory. Toxicity and organ-specific safety signals remain high-risk, high-cost challenges. Models such as Ncardia’s Heart in a Box offer a high-throughput approach to studying cardiotoxicity, with human biology at the center. Full replacement of animal studies is unlikely in the near term, but selective replacement is gaining ground. Regulatory momentum and public pressure also play a role. Both point toward a future where human-relevant data carries greater weight.
He also expects deeper use of patient-derived iPSCs as the field matures. Precision disease models and stratified assays will push discovery into more nuanced territory. Artificial intelligence will accelerate hypothesis generation, but the same bottleneck the field faces today will remain. AI cannot outperform the data it receives. The need for large, standardised, high-quality datasets will only grow.
Through all of this, de Groot returns to a principle he learned early in his bench science days. You start with the biology and with the patient endpoint. Everything else flows from there.
“Science rarely moves in a straight line,” he said.“Structured risk management is not a constraint. It is how you protect outcomes.”
For leaders weighing whether to adopt iPSC-based platforms, his advice is direct. Begin now. Experience matters. Internal datasets compound. The technology does not need to be flawless to shift decision quality in the right direction. Waiting carries its own cost. Those who build capability today will be better positioned for the emerging standards.
In a field defined by complexity and uncertainty, the search for human relevance has become one of the clearest paths forward. De Groot and his team believe the tools are finally mature enough to make that path more accessible. The companies that follow it stand to make faster, better, and more biology-relevant decisions.