What Effective CDMO Oversight Looks Like
Tags :
Biopharma
Share :
07.11.2026
A capable CDMO can manufacture a product well and still be part of an oversight arrangement that leaves important decisions unresolved. The manufacturer may understand its process and equipment better than the company commissioning the work. That company may hold development history or regulatory commitments that the site has not seen. Each can reach a reasonable conclusion from an incomplete picture.
Effective CDMO oversight gives both parties the information and authority they need to control the product. It extends beyond establishing that a site is suitable. It has to make the relationship workable when the evidence is incomplete, a change affects one product differently from the rest of a platform, or the original supply plan can no longer be supported.
The oversight model therefore belongs within the pharmaceutical quality system, with practical consequences for resourcing and CMC strategy. An annual audit and quarterly review may be appropriate, but their frequency says little about whether a significant issue will reach the right people in time.
EU GMP Chapter 7 places the control and review of outsourced activities within the contract giver’s pharmaceutical quality system. It also addresses the knowledge that the contract giver provides to the manufacturer. Oversight depends on the quality of that exchange in both directions.
The sponsor does not need to recreate the CDMO’s laboratory or duplicate every specialist role. It needs enough product and process knowledge to assess conclusions that affect its product. That includes the rationale behind the control strategy and the limits of the evidence supporting it.
A technically experienced manufacturer cannot fill a gap in product history that has never been shared.
A hold-time limit illustrates the point. A transfer package may record the approved limit without explaining why it was selected, where the supporting data were limited or what follow-up work was expected. When a later change is proposed, the number alone may not be enough to judge its significance. Preserving the reasoning matters, particularly after the original development team has moved on.
The same applies when evidence is distributed across organisations. A drug substance investigation may need to be read alongside later drug product performance, stability results or complaints held elsewhere. Individual reports can be adequate within their scope while nobody has assessed the combined implications. The sponsor needs a way to bring that evidence together and record what it means for the product.
That work requires available expertise. A responsibility matrix that assigns a review to QA does not establish who can assess a complex analytical investigation, or whether that person has time reserved for it. The oversight workload belongs in the outsourcing budget. External expertise can be appropriate for a small biotech, provided that product knowledge is retained between assignments rather than rebuilt during each urgent investigation.
GMP Bridge’s clinical-stage biotech case study illustrates the connection between manufacturing strategy and the oversight model. The engagement combined CDMO evaluation with governance and operational review routines for batch release, deviations and changes. Those arrangements formed part of the manufacturing strategy itself.
FDA’s final guidance on quality agreements makes clear that neither party can delegate its applicable statutory or regulatory CGMP responsibilities through an agreement. The guidance contains nonbinding recommendations for commercial manufacturing and recognises that many of its principles can also be useful before commercialisation. The agreement allocates activities; each party retains the responsibilities that apply to its role.
A useful test is a proposed change on which the parties could reasonably disagree. For example, a CDMO might propose a new filter supplier across a manufacturing platform. The site may have evidence of an operational benefit, while one client needs additional product-specific assessment of comparability or the validated state. The agreement should make clear how that assessment will happen, which changes need prior approval and who determines the regulatory implications.
Timing determines whether those provisions are usable. Where the sponsor’s input is needed before implementation or a disposition decision, the review period should allow for the evidence to be assessed and questions resolved. Other categories of change may legitimately be notified afterwards. Those categories need to be agreed deliberately, consistent with applicable requirements and product commitments, rather than left to different interpretations at each organisation.
Escalation needs similar care. Naming two senior contacts does not resolve a technical disagreement unless they have access to the evidence and know how the issue will be assessed. The arrangement should address what remains on hold while a significant uncertainty is unresolved. Commercial escalation must not override the responsibilities of the Quality Unit or any applicable Qualified Person certification duties.
Access to information should be proportionate to the decision. A well-supported summary may be sufficient for some reviews. Others require underlying records or a discussion with the people who generated the data. The practical arrangements for that access, including protection of other clients’ confidential information, should be established before an urgent product decision depends on them.
This also requires judgement from the sponsor. Demanding every record for every event can consume the time needed to examine the few issues that materially affect the product. The depth of review should follow the risk and the uncertainty, with a clear explanation for when a closer assessment is needed.
Performance reviews should show whether the relationship is maintaining control over time. Investigation closure dates and on-time delivery figures are useful, but can conceal recurrence or a technical question that remains unresolved. Several minor deviations with plausible individual explanations may, when read together, suggest an ineffective CAPA or a process that is drifting.
The review needs to connect those events, not merely count them.
The sponsor’s own performance belongs in that assessment. If an investigation has waited three weeks for its review, the delay should remain visible rather than disappear into the CDMO’s cycle-time figures. Looking at both sides helps distinguish incomplete evidence from unavailable expertise or an unclear decision route. Adding another meeting is unlikely to fix any of those problems on its own.
The focus should evolve as the product develops. For an early clinical programme, oversight needs to address product-specific uncertainties even where the CDMO operates an established platform. A small number of batches does not remove the need for appropriate GMP control, but it may justify a different allocation of expertise and review effort from a mature commercial product.
As commercial supply approaches, the model should be revisited alongside process validation and inspection readiness. Arrangements that supported clinical batches may need to change for routine release, continued process verification and post-approval changes across several markets. The relevant organisations should be able to explain their respective decisions and retrieve the evidence supporting them. Reliance on a long-standing working relationship is insufficient if that knowledge sits with one person or cannot be demonstrated.
Changes in the external manufacturing network also call for a product-specific response. A new subcontracted laboratory requires the relevant prior evaluation and approval and may warrant an audit. A change in legal ownership without operational consequences may require a different assessment. The question is what has changed for the product and the arrangements that control it, rather than whether the original supplier qualification file is still complete.
For an established relationship, a recent investigation or change can provide a more revealing review than another general supplier questionnaire. Reconstructing the event shows when each party learned about it, what evidence was available and whether the people responsible could act before the outcome was effectively fixed. It can reveal a narrow weakness, such as an access provision that did not work or an expert who was never available, as readily as a broader governance problem.
A sound arrangement should withstand that review without relying on the same individuals to rescue it each time. Technical differences will still occur. Effective oversight makes it possible to examine them properly, reach a defensible decision and preserve the reasoning for the next time the product changes.
GMP Bridge supports sponsors with CDMO quality oversight and supplier control, from partner evaluation to the governance and review practices needed in an existing manufacturing relationship.
For a monthly perspective that connects the dots in biopharma Quality, request access to the GMP Bridge Briefing.
Tags :
Share :
European Commission. EudraLex Volume 4, EU GMP Part I, Chapter 7, Outsourced Activities. In operation since 31 January 2013. See sections 7.4 to 7.8, 7.11 and 7.14 to 7.17.
US FDA. Contract Manufacturing Arrangements for Drugs: Quality Agreements. Guidance for Industry, final, November 2016. Nonbinding recommendations.
Disclaimer:
In this article, “sponsor” means the company commissioning the outsourced work. The duties of a clinical trial sponsor, product owner, marketing authorisation holder and manufacturing authorisation holder are not interchangeable. Applicable responsibilities depend on the product, development stage and market.