EU regulators confront the evidence gap for drug-device combinations
The rapid rise of medicinal products developed alongside medical devices, in vitro diagnostics, software, and companion diagnostics has exposed a structural problem in EU development pathways: the evidence needed to support each component is often reviewed in separate regulatory lanes. For sponsors, that can mean generating clinical, analytical, and performance data under one framework while trying to satisfy a different set of expectations for the paired product. For patients, the consequence is slower translation of integrated technologies that depend on the medicine and device functioning as a single therapeutic system.
A dedicated Focus group was convened after the 9th EMA-Industry stakeholder platform on research and development support in December 2022. Its mandate was practical and tightly defined: determine whether scientific advice could be delivered in a multistakeholder setting for medicinal products used in combination with a medical device or an IVD, including companion diagnostics. The group brought together the EMA Scientific Advice Working Party, national competent authorities responsible for medicinal products, authorities overseeing medical devices, notified bodies, industry, and the EMA itself. The central question was not whether these products deserve advice, but how that advice should be structured so the evidence plan matches the combined product from the outset.
The need is easy to understand at the bedside. A therapy paired with a diagnostic does not work as two separate regulatory objects; it works as a linked decision system. If the diagnostic misclassifies patients, the medicine may appear ineffective or unsafe. If the device changes how the medicine is selected, dosed, or monitored, the performance characteristics of the device become inseparable from the drug’s clinical value.
Nine case studies exposed where separate review pathways break down
The Focus group examined nine study cases: four drug-device combinations and five drug-IVD or drug-companion diagnostic combinations. Stand-alone high-risk devices and stand-alone IVDs were excluded because a separate scientific advice pilot already exists for medical device expert panels. Low-risk stand-alone device and IVD developments were also outside scope. The point of the exercise was narrower and more demanding: identify the questions that only become answerable when drug, device, diagnostic, and regulatory experts are in the same room.
Across those nine cases, the group explored what kinds of scientific issues benefit from comprehensive discussion during evidence generation planning. The cases were used to test how a multistakeholder process might work in practice, and which decision makers and specialists would need to participate for the advice to be credible and actionable. The discussion was not limited to procedural coordination. It dealt with the deeper scientific consequences of pairing a medicine with a device or diagnostic: what should be measured, in whom, at what stage of development, and how the evidence package should be balanced between the components.
Some combinations require synchronized validation. If the companion diagnostic is intended to select the subgroup most likely to benefit, the drug development plan cannot be designed sensibly without understanding the diagnostic’s intended use, analytical performance, and regulatory trajectory. The same logic applies to software-driven or hardware-assisted administration systems, where the delivery tool may influence dose precision, adherence, interpretation of results, or downstream safety monitoring.
Why a joint scientific advice model matters for development planning
The regulatory challenge is not merely administrative. Medicinal products, medical devices, and IVDs sit under different EU frameworks, each with its own evidentiary expectations, technical language, and decision-making authorities. A sponsor trying to develop a combined product may receive scientifically reasonable but incomplete advice if each component is assessed in isolation. One authority may focus on clinical efficacy, another on diagnostic performance, and another on conformity or device-specific risk. Without a shared forum, the development plan can drift toward fragmented endpoints, duplicative studies, or avoidable delays.
A multistakeholder advice model would allow scientific expectations to be aligned before key studies begin. That is especially important for evidence generation planning, where early design choices determine whether a development program can later support both market authorization and safe clinical use. For companion diagnostics, the value is obvious: the therapeutic claim depends on correct patient identification. For device-supported drug delivery or monitoring, the device’s reliability may shape the medicine’s real-world benefit just as much as the pharmacology does.
The Focus group’s discussions also point to a broader policy need. Innovative healthcare products increasingly blur the boundary between therapy, measurement, and software-assisted decision-making. Regulatory systems built around separate categories can manage that complexity only if they create a mechanism for joint scientific questioning. Without it, sponsors may face inconsistent advice, and regulators may be forced to evaluate parts of the same therapeutic proposition in isolation.
The policy question now is how to operationalize cross-framework advice
The article presents the group’s insights as a foundation for future policy and technical discussions rather than as a final operating model. Even so, the direction is clear: the EU needs a structured way to decide which combined-product questions merit joint scientific advice, who should attend those discussions, and how the resulting recommendations should be translated across regulatory frameworks. Those decisions will matter for sponsors developing novel integrated products, but they will also shape how efficiently patients gain access to therapies whose success depends on both a medicine and a companion technology.
The work leaves several operational issues open. The article does not define a finalized process for selection of cases, the exact division of responsibility among regulators and notified bodies, or how advice would be harmonized when component-specific requirements differ. Those unanswered implementation details will determine whether the model becomes a useful development tool or remains a conceptual bridge between frameworks. For now, the clearest result is that the science of combined products cannot be reviewed product by product alone; the evidence package has to be built around the interaction between them from the start.
- • Frontiers in Medicine 2025 DOI: https://doi.org/10.3389/fmed.2025.1593644