MDCG 2025-5: IVD performance studies - Q&A, explained

Who this applies to

This guidance is for sponsors of IVD performance studies, including manufacturers, universities, hospitals and other non-commercial sponsors. It is also relevant to investigators, ethics committees, competent authorities and teams preparing submissions under IVDR Articles 57–77.

The sponsor label matters. A study does not fall outside the IVDR because it is academically initiated, unfunded or not intended to support an immediate CE-marking application. If it is designed to establish or confirm analytical or clinical performance of an IVD, it is a performance study under the IVDR.

Which studies need an application or notification?

All IVD performance studies must meet the applicable Article 57 requirements, but not all are submitted to a competent authority. The route depends mainly on participant procedures, clinical use of the result and whether a CE-marked IVD stays within its intended purpose.

Blood draws, punctures and fresh-tissue biopsies can be surgically invasive sample-taking. The words “only for the study” are important: using part of a sample already taken for routine care is different from asking for a new procedure solely to generate study material.

“Interventional” is also broader than physically intervening. If an IVD result can influence diagnosis, treatment or patient management, false positive and false negative results can expose the subject to risk. Blinding the result from clinical decision-makers can therefore change the submission analysis.

Left-over samples are genuinely left over

MDCG 2025-5 draws a firm line around left-over samples. They were collected previously for another purpose and remain after that purpose has been fulfilled. Samples collected prospectively for the performance study are not left over just because the collection would also occur in normal care.

That distinction is especially important for companion diagnostics. A left-over-sample-only CDx study follows the notification route; other CDx performance studies require an application.

Submission and study conduct

Applications should use the documentation structure in MDCG 2022-19. Until the relevant Eudamed functionality is available, sponsors must follow national submission channels. A multinational study needs the required submission in every Member State where subjects participate; one national approval does not cover the Union.

The guidance also clarifies several operational points:

• One performance study plan may cover multiple IVDs if each device, objective and analysis is unambiguously documented.
• The instructions and participant-facing information must comply with national language requirements and be understandable to their intended readers.
• For a prospective study, “start” is the first act of recruitment. For a left-over-sample-only study, it is the first study-specific analysis of samples.
• A study requiring authorisation cannot begin without it. Notification pathways do not produce the same explicit authorisation, but their statutory periods and national checks still have to run.
• Full technical documentation is not routinely filed with every application, but the sponsor must provide enough evidence of conformity with the applicable general safety and performance requirements and make further documentation available on request.

Substantial modifications

The sponsor decides initially whether a modification is substantial, but the competent authority can disagree. The test is whether the change is likely to have a substantial impact on subject safety, rights, or the reliability and robustness of the data.

The guidance's appendix gives 29 examples. Changes to core objectives, key endpoints, subject population, sample-taking, risk controls, statistical assumptions or result use are much more likely to be substantial than administrative corrections. Where Article 71 applies, submit the updated documents and do not implement the change until the applicable review period has passed or earlier authorisation is given. The guidance explains the normal 38-calendar-day period and the possible expert-consultation extension.

What this means for you, practically

1. Classify the study before writing the submission. Document whether sampling is study-only and surgically invasive, whether results reach clinical decision-makers, whether there are additional procedures or risks, and whether it is a CDx study.
2. Prove sample provenance. Your protocol and sample log should show when, why and under which consent the material was collected. Do not use “left over” as a convenient label for prospective collection.
3. Map every Member State. Build the national authority, ethics, language, fee and timeline requirements into the project plan before opening sites.
4. Control the real study start. Recruitment and left-over-sample analysis have different start events; configure sites and laboratories so neither occurs early.
5. Run amendments through a documented substantial-modification assessment. Use the appendix examples, record the rationale, and hold implementation where notification or authorisation is required.