Article
CMC reality check: Late-stage risks that impact approval
-
Martin Mewies, PhD
-
Thomas de Marchin, PhD
-
Michael Day, PhD
Understanding the key bottlenecks in late-stage CMC development and how they impact submission success
Biopharmaceutical companies face several common CMC (chemistry, manufacturing, and controls) challenges, which typically emerge in late-stage development and weigh heavily on submission success. Key among these are specification justification, process validation, comparability, and manufacturing readiness.
At the center of these issues is the overall control strategy: whether the applicant can demonstrate an adequate understanding of raw materials, critical process parameters, critical quality attributes, analytical methods, and manufacturing variability, and show how those elements collectively assure consistent product quality.
Many developers underestimate what is required to achieve manufacturing readiness. If regulators have any doubts about product quality, manufacturing process consistency, or the reliability of clinical trial data, it may trigger an inspection and can result in approval delays – or outright rejections of the application.
At the center of these issues is the overall control strategy: whether the applicant can demonstrate an adequate understanding of raw materials, critical process parameters, critical quality attributes, analytical methods, and manufacturing variability, and show how those elements collectively assure consistent product quality.
Many developers underestimate what is required to achieve manufacturing readiness. If regulators have any doubts about product quality, manufacturing process consistency, or the reliability of clinical trial data, it may trigger an inspection and can result in approval delays – or outright rejections of the application.
To navigate these bottlenecks, manufacturers will need to carefully consider early decisions and how to align CMC strategy with regulatory expectations to avoid delays and support successful first-cycle approval.
Key bottlenecks in late-stage CMC development
An issue that is commonly picked up during a gap analysis is non-compendial specifications, particularly around microbiological control that diverges from pharmacopeial methods. An example is bioburden testing for biologics that diverges either for a process reason or a product-related reason.
This often occurs when a small or medium company is under financial pressure to move forward quickly and proposes specifications that are set by a “rule of thumb” rather than being grounded in safety, efficacy, or process capability. As such, they lack the data needed to justify the specification. This makes it very challenging when the manufacturer – or a company that has acquired the product – moves into commercial manufacturing.
Another root cause is where a product is being developed for a particular market and then, as part of global expansion, the manufacturer looks to take that product to other jurisdictions, at which point small divergences from pharmacopeial methods start to become a problem.
A third issue often observed is with potency specifications for legacy products that no longer align with the tightened expectations of the European Medicines Agency (EMA) or the US Food and Drug Administration (FDA). In such situations, manufacturers may seek to retrospectively bring a specification into line with current expectations. This can be particularly challenging, as modifying the manufacturing process at a late stage can be extremely costly in both time and effort. Developers may also lack the statistical expertise needed to establish and robustly justify an appropriate specification.
This often occurs when a small or medium company is under financial pressure to move forward quickly and proposes specifications that are set by a “rule of thumb” rather than being grounded in safety, efficacy, or process capability. As such, they lack the data needed to justify the specification. This makes it very challenging when the manufacturer – or a company that has acquired the product – moves into commercial manufacturing.
Another root cause is where a product is being developed for a particular market and then, as part of global expansion, the manufacturer looks to take that product to other jurisdictions, at which point small divergences from pharmacopeial methods start to become a problem.
A third issue often observed is with potency specifications for legacy products that no longer align with the tightened expectations of the European Medicines Agency (EMA) or the US Food and Drug Administration (FDA). In such situations, manufacturers may seek to retrospectively bring a specification into line with current expectations. This can be particularly challenging, as modifying the manufacturing process at a late stage can be extremely costly in both time and effort. Developers may also lack the statistical expertise needed to establish and robustly justify an appropriate specification.
It is likely that at some point in the product lifecycle there will be a change of manufacturing site, a need to scale up, or, potentially, a change of ownership. Each of these scenarios will necessitate a comparability exercise to demonstrate that the changes in question do not compromise product safety or efficacy.
Unfortunately, gaps in historical data often make it difficult to establish a credible bridge between the old site and the new site, a task that is made more complex if there are process or analytical changes over time. These gaps often become apparent when a large company acquires the assets of smaller developers and needs to shift the production lines, only to find it difficult to reconcile the specifications with manufacturing reality.
A robust comparability exercise also requires clearly defined acceptance criteria. In practice, some degree of difference will always exist following a change, whether attributable to the change itself or to the natural variability inherent in both the process and the measuring equipment, particularly when data are limited. Establishing appropriate equivalence margins and selecting a sound statistical approach to demonstrate comparability are therefore among the most demanding aspects of the exercise.
For many companies, particularly smaller developers that are reliant on contract manufacturing organizations (CMOs), poor management of the process and incomplete documentation can be hugely problematic. If unresolved out-of-specification results or deviations are not properly recorded or the root cause is not identified, that can have a significant knock-on effect later and will be difficult to explain to regulators, reducing confidence.
Unfortunately, gaps in historical data often make it difficult to establish a credible bridge between the old site and the new site, a task that is made more complex if there are process or analytical changes over time. These gaps often become apparent when a large company acquires the assets of smaller developers and needs to shift the production lines, only to find it difficult to reconcile the specifications with manufacturing reality.
A robust comparability exercise also requires clearly defined acceptance criteria. In practice, some degree of difference will always exist following a change, whether attributable to the change itself or to the natural variability inherent in both the process and the measuring equipment, particularly when data are limited. Establishing appropriate equivalence margins and selecting a sound statistical approach to demonstrate comparability are therefore among the most demanding aspects of the exercise.
For many companies, particularly smaller developers that are reliant on contract manufacturing organizations (CMOs), poor management of the process and incomplete documentation can be hugely problematic. If unresolved out-of-specification results or deviations are not properly recorded or the root cause is not identified, that can have a significant knock-on effect later and will be difficult to explain to regulators, reducing confidence.
Another frequent late-stage challenge is discovering that analytical methods that were adequate during development are not sufficiently robust for commercial release, stability testing, or method transfer. Method validation, qualification, transfer, reference-standard strategy, and lifecycle management therefore need to be considered well before submission.
Compounding this, evolving regulatory expectations around quality by design (QbD), alongside increasingly rigorous confidence requirements, have raised the bar for experimental design and the statistical frameworks used to analyze the data.
Stability programs can present similar problems. Insufficient time points, changes in analytical methodology during development, inadequate comparability between development and commercial batches, or an insufficient number of representative commercial-scale batches can limit the shelf life that regulators are prepared to accept at approval.
These shortcomings are particularly difficult to remediate late in the process, as the missing data often cannot be reconstructed or extrapolated and may require additional long-term stability data to be generated. This can result in costly and avoidable delays.
Compounding this, evolving regulatory expectations around quality by design (QbD), alongside increasingly rigorous confidence requirements, have raised the bar for experimental design and the statistical frameworks used to analyze the data.
Stability programs can present similar problems. Insufficient time points, changes in analytical methodology during development, inadequate comparability between development and commercial batches, or an insufficient number of representative commercial-scale batches can limit the shelf life that regulators are prepared to accept at approval.
These shortcomings are particularly difficult to remediate late in the process, as the missing data often cannot be reconstructed or extrapolated and may require additional long-term stability data to be generated. This can result in costly and avoidable delays.
An issue that commonly arises is process validation and an over-reliance on the traditional “three validation batches” as a default. Modern lifecycle approaches to process validation expect the extent of process performance qualification (PPQ) to be scientifically based and risk-based, considering process knowledge, variability, manufacturing experience, scale, and the ability of the PPQ program to provide adequate assurance of continued process performance.
Statistical approaches can play an important role in demonstrating that assurance rather than simply relying on a predetermined number of batches.
Inadequate or poorly designed experiments will impact knowledge about the product and robustness of the CMC process, which will result in validation issues later in development.
Statistical approaches can play an important role in demonstrating that assurance rather than simply relying on a predetermined number of batches.
Inadequate or poorly designed experiments will impact knowledge about the product and robustness of the CMC process, which will result in validation issues later in development.
A practical approach to managing bottlenecks
Many of the bottlenecks that occur do so due to short-term thinking, poor data governance, lack of oversight of CROs or the manufacturing process generally, and narrowly focused due diligence processes.
Manufacturers can reduce the risk of late-stage problems arising with early planning and adopting a risk-based approach to the manufacturing process. For example, divergences in pharmacopeial methods can be mitigated by assessing alignment across all intended markets from the outset rather than as an afterthought during global expansion.
It is imperative that manufacturers build product knowledge from the earliest possible stage and understand and document the critical quality attributes. Manufacturers need to invest in sufficient tests to understand process capability and potential risks, and set meaningful, justifiable specifications.
At the same time, by applying QbD thinking from the outset, manufacturers will be able to achieve greater, more robust process knowledge, often with fewer experiments. Proper data structure (for example, sampling at informative time points in stability studies) can make 100 well-designed data points more powerful than 1,000 poorly collected ones.
Manufacturers can reduce the risk of late-stage problems arising with early planning and adopting a risk-based approach to the manufacturing process. For example, divergences in pharmacopeial methods can be mitigated by assessing alignment across all intended markets from the outset rather than as an afterthought during global expansion.
It is imperative that manufacturers build product knowledge from the earliest possible stage and understand and document the critical quality attributes. Manufacturers need to invest in sufficient tests to understand process capability and potential risks, and set meaningful, justifiable specifications.
At the same time, by applying QbD thinking from the outset, manufacturers will be able to achieve greater, more robust process knowledge, often with fewer experiments. Proper data structure (for example, sampling at informative time points in stability studies) can make 100 well-designed data points more powerful than 1,000 poorly collected ones.
Oversight and expert engagement
Proper oversight of stakeholders and of data is imperative. When working with CMOs, manufacturers need to establish clear contractual requirements for complete data packages, timely deviation documentation, and root cause analysis of anomalies. This means treating CMO data governance as a CMC risk and requirement.
Manufacturers can also help to mitigate later issues by bringing in regulatory and CMC experts to conduct early gap analysis and to identify where specifications may need to be tightened or justified with agencies.
Manufacturers can also help to mitigate later issues by bringing in regulatory and CMC experts to conduct early gap analysis and to identify where specifications may need to be tightened or justified with agencies.
Additionally, to help ensure data gathered is of sufficient quality, manufacturers should consider bringing in statisticians early to guide how data should be collected rather than being brought in after the fact simply to analyze the data. Having good quality data from the outset makes it easier to:
- Propose and defend specifications that are both regulatory-compliant and manufacturable
- Scientifically and statistically justify decisions
- Build credible comparability bridges using historical data, avoiding the need to regenerate batches at sender sites
- Demonstrate manufacturing capability clearly during due diligence or, where relevant, acquisition review
Conclusion: A philosophy of data integrity
What underpins all CMC activities, and will be vital for a sound regulatory submission, is the availability of adequate and appropriate data generated from the earliest stages of development. To prevent and overcome the CMC issues that often occur later in development, manufacturers should treat data integrity as a manufacturing philosophy, not merely a compliance checkbox.
About the authors:
Martin Mewies, PhD, is Director, Regulatory Affairs, CMC, at Cencora. He has over 30 years of experience in protein biochemistry, with more than 20 years in biologics CMC/regulatory.
Thomas de Marchin, PhD, is Associate Director of Statistics and Data Science at Cencora, where he applies advanced statistical methodologies and machine learning algorithms to optimize drug discovery and manufacturing efficiency. With deep expertise in regulatory compliance – particularly FDA and GMP standards – he bridges the gap between complex analytical approaches and practical pharmaceutical applications.
Michael Day, PhD, is Senior Director of Regulatory Strategy and CMC at Cencora. He brings more than 25 years of regulatory, CMC, and quality experience, including direct interactions with FDA on CRLs and BLA submissions in cell and gene therapies.
Thomas de Marchin, PhD, is Associate Director of Statistics and Data Science at Cencora, where he applies advanced statistical methodologies and machine learning algorithms to optimize drug discovery and manufacturing efficiency. With deep expertise in regulatory compliance – particularly FDA and GMP standards – he bridges the gap between complex analytical approaches and practical pharmaceutical applications.
Michael Day, PhD, is Senior Director of Regulatory Strategy and CMC at Cencora. He brings more than 25 years of regulatory, CMC, and quality experience, including direct interactions with FDA on CRLs and BLA submissions in cell and gene therapies.
Clause de non responsabilité :
Les informations fournies dans cet article ne constituent pas des conseils juridiques. Cencora, Inc. encourage vivement les lecteurs à consulter les informations disponibles relatives aux sujets abordés et à s’appuyer sur leur propre expérience et expertise pour prendre des décisions à ce sujet.
