Insights: January 2025

January 2025’s Clinical Quality Reset: ICH E6(R3), FDA AI Credibility, and the CTIS Deadline

January 2025 did not deliver incremental regulatory change. It delivered a reset. Within the first week of the month, ICH E6(R3) reached Step 4, and FDA issued its first draft guidance on the use of AI to support regulatory decision-making for drugs and biologics. By month-end, the EU’s CTIS transition period had ended, bringing all ongoing EU trials fully under the Clinical Trials Regulation. For biotech sponsors, those are not isolated developments. They are a clear signal that quality, oversight, and regulatory credibility are being pulled earlier into development and made far more explicit.

From a quality leadership perspective, the common thread is straightforward: regulators are asking sponsors to demonstrate that they understand what matters most, apply controls in proportion to risk, and can defend the reliability of their methods, systems, and decisions. Whether the issue is protocol design, outsourced trial activity, AI-generated evidence, or EU trial submissions, the era of loosely connected compliance activities is ending.

ICH E6(R3): quality is no longer something you inspect in later

ICH E6(R3), adopted on 6 January 2025, makes the direction unmistakable. The guideline embeds quality by design, requires prospective identification of factors critical to quality, expects trial processes to be proportionate to participant risk and the importance of the data collected, and states that information and systems used in clinical trials should be fit for purpose. It also makes sponsor quality management more explicit, including the expectation that the sponsor describe the quality management approach used in the trial in the clinical study report.

That matters because many organizations still behave as though GCP quality is primarily a downstream audit function. E6(R3) points in the opposite direction. Sponsors retain ultimate responsibility for trial-related activities that are transferred to service providers, must ensure appropriate oversight of important outsourced and subcontracted work, and are expected to maintain computerized systems that are validated, secure, and fit for purpose based on risk. In practical terms, this elevates QA into protocol design, operational feasibility, vendor governance, data strategy, and digital system oversight rather than leaving it concentrated in inspection preparation and CAPA management.

For biotech companies, this is the real GCP reset. The strongest quality organizations in 2025 will not be the ones with the largest audit schedules. They will be the ones that can define their critical-to-quality factors early, build proportionate controls around them, and show that the sponsor—not the CRO, not the platform vendor, not the software provider—remains firmly in command of trial quality.

FDA’s AI draft guidance: “credibility” is now a regulatory deliverable

FDA’s January 2025 draft guidance is just as important, even for companies that still describe their AI work as exploratory. Issued on 6 January 2025, the guidance is FDA’s first for AI used in the development of drug and biological products. It introduces a risk-based, 7-step credibility assessment framework built around a defined question of interest, a specific context of use, model risk, a plan to establish credibility, execution of that plan, documentation of the results, and a final determination of whether the model is adequate for its intended use.

The most important point for sponsors is that credibility is not a general claim about how advanced a model is. Under FDA’s framework, credibility is tied to a specific context of use, and the rigor of the supporting evidence must be commensurate with model risk. FDA defines model risk through two dimensions: model influence and decision consequence. The agency also strongly encourages early engagement, expects a documented credibility assessment plan, and contemplates a credibility assessment report that may become part of a submission or be made available on request.

That has immediate implications for biotech quality and technical governance. AI models that support development, manufacturing, or regulatory decisions cannot sit outside the quality system as innovation projects with light documentation. The draft guidance discusses life cycle maintenance, ongoing performance monitoring, and the need to manage issues such as performance change over time and data drift. FDA also notes that the guidance reflects real review experience, including more than 500 drug and biological product submissions with AI components since 2016. The message is clear: if AI will influence a regulated decision, its governance must be inspection-ready and submission-ready.

CTIS: January closed the grace period

On the EU side, January ended the transition period and removed any remaining ambiguity. EMA stated that from 31 January 2025, all clinical trials in the EU, including ongoing trials originally approved under the Clinical Trials Directive, are governed by the Clinical Trials Regulation. EMA had already warned sponsors a year earlier that transitions could require up to three months for Member State authorization, and by 31 January 2025 it confirmed that remaining ongoing trials not moved into the new system may be subject to corrective measures. Transition procedures are no longer available, and sponsors of ongoing CTD trials are required to submit a new application via CTIS.

Quality leaders should not view this as a narrow regulatory-operations milestone. CTIS is the single entry point for submission, assessment, and oversight of EU clinical trials, and it includes a public searchable database. That makes portfolio governance, document consistency, role clarity, and submission discipline far more visible than they were under the previous framework. Before January closed, sponsors needed more than a filing plan; they needed a sponsor-level view of which legacy studies were still active, whether their transition path was realistic, and whether their internal ownership model was strong enough to support both compliance and transparency.

What biotech QA leaders should do with this signal

1. Move quality upstream. Define critical-to-quality factors earlier, build proportionate controls around them, and use pre-specified ranges such as trial-level quality tolerance limits where relevant instead of waiting for downstream findings to reveal design weaknesses.
2. Put AI under formal governance before it reaches a submission. Require a defined context of use, model risk assessment, credibility plan, documented results, and life cycle monitoring before AI-derived outputs are allowed to influence regulated decisions.
3. Treat CTIS as executive portfolio governance, not just study start-up mechanics. Every EU trial should have a clear regulatory status, ownership path, and operating model for what is submitted, maintained, and publicly disclosed.
4. Re-underwrite sponsor oversight of vendors and systems. E6(R3) is explicit that outsourced activity does not outsource accountability, and that sponsor oversight must extend to important transferred activities, including subcontracted work and fit-for-purpose computerized systems.

January 2025 will likely be remembered as the month clinical quality stopped being discussed as a support function and re-emerged as a strategic operating discipline. E6(R3) raised the bar for designed-in quality and sponsor control. FDA translated AI “credibility” into a structured regulatory expectation. CTIS closed the door on transition and reinforced that transparency and execution discipline are now inseparable in Europe. The biotech companies that respond fastest will not just be more compliant. They will be better governed, easier to defend, and better positioned to scale development without losing control of risk.