A Release-by-Exception Model for Cell Therapy Quality Decisions

Key Takeaways

• Autologous cell therapy batch release decisions depend on predefined specifications and written quality procedures, not ad hoc testing outcomes.

• Regulatory expectations require acceptance criteria that incorporate appropriate statistical quality control for product release.

• Potency testing for cellular and gene therapy products is an essential part of release justification under current FDA guidance.

• Quality risk management principles from ICH Q9 provide the foundation for structured exception evaluation in release decisions.

• A release-by-exception approach strengthens documentation, traceability, and regulatory confidence within the pharmaceutical quality system.

For autologous cell therapies, batch release is not a routine manufacturing checkpoint but a time-critical quality decision tied to an individual patient. Each lot is patient-specific, inherently variable, and often subject to narrow clinical windows that limit the opportunity for repeat testing or extended hold times. These characteristics distinguish autologous products from traditional large-batch biologics and place unusual pressure on quality control (QC) and quality assurance functions to reconcile scientific uncertainty with regulatory accountability.

Regulatory frameworks make clear that release is grounded in predefined specifications and written sampling and testing procedures, supported by appropriate statistical QC principles.¹ For biological products, sterility and potency occupy a central role in this decision process, with sterility testing required for each lot and potency expectations tied to demonstration of product quality and clinical relevance.^2,3 At the same time, quality risk management and pharmaceutical quality system (PQS) principles emphasize that decisions affecting product quality must be made using structured, risk-based approaches across the product life cycle.^4,5

These overlapping demands mean that batch release cannot be reduced to a single pass/fail laboratory outcome. Instead, it requires integration of predefined acceptance criteria, constrained test results, and broader product and process knowledge within a governed quality system. Process analytical technology (PAT) and real-time data concepts provide additional context for how manufacturing and in-process information can contribute to release decisions, but they operate within the same regulatory structure that governs traditional testing rather than replacing it.⁶

In this setting, analytics-driven batch release is best understood not as automated approval, but as a structured release-by-exception framework embedded within the PQS. Under this model, routine batches proceed through established specifications and procedures, while analytics and quality governance are applied when results fall outside expected ranges or when additional evidence is required to justify a decision. This framing shifts the focus from individual test results to disciplined decision-making, preserving regulatory responsibility while enabling more consistent and transparent use of data in autologous cell therapy manufacturing.

Batch release decisions for autologous cell therapies are grounded in long-standing regulatory expectations that define how products are tested, evaluated, and approved for distribution. These requirements establish release not as an informal judgment call, but as a governed process based on predefined criteria, documented procedures, and quality system oversight. Together, they create the structural basis for any analytics-driven approach to release decision making.

Regulations governing testing and release make clear that each batch must be evaluated against predefined specifications using written sampling and testing procedures with defined acceptance and rejection criteria.¹ These procedures are not optional or ad hoc; they must be established in advance and followed consistently for every lot. Importantly, the same regulation explicitly links release decisions to appropriate statistical QC principles, reinforcing that acceptance criteria should be scientifically and quantitatively justified rather than based solely on qualitative judgment.

Within this framework, release is not triggered by a single data point but by conformance to a structured set of expectations that combine analytical results with procedural discipline. This regulatory foundation is what allows analytics to play a role in release decisions without displacing the underlying authority of specifications and written procedures.

For biological products, sterility and potency are central elements of release. Regulations require sterility testing of each lot of a biological product’s final container material or other approved material, making sterility a mandatory gate for release.² At the same time, potency testing for cellular and gene therapy products is expected to demonstrate that the product’s biological activity is consistent with its intended clinical effect and aligned with data generated during development.³

These requirements introduce challenges for short-lived, patient-specific products. Sterility tests rely on incubation periods that extend over days, while potency assays may be complex, variable, and difficult to complete within narrow clinical timelines. As a result, sponsors must reconcile the obligation to meet sterility and potency expectations with the practical realities of delivering autologous therapies to patients in a timely manner. This tension is not a deviation from regulatory principles but an inherent feature of applying them to highly individualized products.

Beyond individual tests, release decisions are situated within the PQS, which serves as the formal authority for how quality-related decisions are made and documented across the product life cycle. Quality risk management principles emphasize that decisions affecting product quality should be based on structured, risk-based approaches that consider both scientific evidence and potential impact on patient safety.⁴ The PQS framework further incorporates ongoing process performance and product quality monitoring as continuous activities rather than isolated events.⁵

Within this governance structure, batch release is one expression of broader life cycle control. Data from manufacturing, testing, and historical performance are evaluated under defined quality procedures, with accountability resting in the quality organization rather than in automated tools or isolated functional groups. PAT and real-time data concepts can support this evaluation, but they remain subordinate to the same quality system expectations that govern conventional release testing.⁶

These combined regulatory and quality system expectations create the need for a structured and defensible way to manage uncertainty when results do not fit neatly within routine release pathways. Rather than weakening regulatory control, this reality reinforces the importance of disciplined decision frameworks that integrate specifications, risk-based evaluation, and documented quality governance.

Traditional batch release models were developed for products manufactured in large, relatively uniform lots, where variability can be absorbed through scale and where retesting or extended investigations are often feasible. Autologous cell therapies operate under fundamentally different conditions. Each batch corresponds to a single patient, and release decisions are therefore inseparable from individual clinical timelines. This patient-specific structure limits opportunities for repeated testing and places heightened emphasis on making defensible decisions with the data available at the time of release.

These constraints are intensified by the central role of sterility and potency in regulatory expectations for biological products. Sterility testing is required for each lot, and established methods rely on incubation periods that extend over days rather than hours.^2,7 Potency expectations for cellular therapies further require evidence that biological activity remains aligned with product quality and clinical relevance, even as assays may be complex and variable.3 When combined with clinical scheduling and short product lifetimes, these requirements create unavoidable time pressure that does not exist to the same degree for conventional biologics.

Within this environment, we can compare two conceptual release models. In a routine release model, batches that meet all predefined acceptance criteria proceed directly to release, and those that do not are rejected. This binary structure assumes that results are unambiguous and that sufficient time exists to resolve discrepancies through additional testing or investigation. In contrast, an exception-based release model recognizes that results may be incomplete, borderline, or difficult to interpret within the available window. Here, batches that do not fit neatly into predefined criteria are routed into a structured evaluation pathway governed by quality risk management principles and PQS oversight.^4,5

The core decision challenge for autologous cell therapies is therefore not whether routine release can be achieved when all criteria are clearly met. It is how to evaluate and justify release when exceptions occur — when results fall outside expected ranges, when supporting data must be weighed alongside final test outcomes, or when time constraints limit the ability to wait for complete information. A release-by-exception framework provides a disciplined way to manage these situations by embedding scientific judgment, risk-based evaluation, and documentation within established quality governance structures rather than relying on ad hoc or informal decision making.

A release-by-exception framework provides a structured way to manage batch release when outcomes do not conform neatly to routine expectations. Rather than treating release as a binary laboratory result, the framework defines how data are evaluated, when exceptions are triggered, and how decisions are governed within the PQS. Its purpose is not to relax specification, but to ensure that departures from routine pathways are handled consistently, transparently, and in alignment with regulatory and quality system principles.

The first step in this framework is to define release criteria and acceptance limits in advance. Specifications and acceptance criteria must be established before manufacturing and testing occur and embedded within written sampling and testing procedures. These limits are tied to critical quality attributes and form the baseline against which every batch is evaluated. By anchoring decisions to predefined criteria, the framework ensures that exception handling does not become subjective or improvised but remains grounded in documented quality expectations.

The second step is to establish authorized data inputs for release decisions. Final QC test results remain central, but they are complemented by manufacturing and in-process data as well as outputs from process performance and product quality monitoring systems. PAT and real-time data streams can contribute to this evidence base, provided they are integrated within the same regulatory structure that governs conventional testing rather than positioned as replacements for it. This step clarifies which data sources are legitimate for supporting release decisions and prevents informal or unvalidated information from influencing outcomes.

The third step is to define what constitutes an exception and how it is triggered. An exception may arise when results fall outside established acceptance criteria, but it may also occur when results technically meet criteria yet display unexpected variability or trends relative to historical performance. Predefining these triggers is essential to maintaining consistency. Without clear thresholds for escalation, exception handling risks becoming reactive or uneven across batches and programs.

The fourth step is to evaluate the exception using risk-based criteria. Once an exception is triggered, the batch enters a structured evaluation pathway governed by quality risk management principles. This evaluation considers the relationship of the observed data to product quality and patient safety, draws on historical process knowledge, and applies documented risk-based reasoning to determine whether sufficient evidence exists to support release.⁴ Throughout this process, decision authority remains with quality assurance, preserving regulatory accountability and ensuring that analytics inform rather than replace human judgment.

The final step is to document, trend, and improve. Each exception decision and its rationale are captured within the PQS, creating a traceable record of how uncertainty was managed. Over time, these records enable trending of exception events and identification of recurring issues, which in turn feed back into process improvement and refinement of acceptance criteria. In this way, release-by-exception becomes not only a control mechanism but also a learning system that strengthens consistency and regulatory confidence across the product life cycle.

Together, these five steps transform batch release from an isolated quality checkpoint into a governed decision framework. By embedding predefined criteria, authorized data inputs, risk-based evaluation, and continuous learning within the PQS, release-by-exception provides a disciplined approach to managing the inherent uncertainty of autologous cell therapy manufacturing while maintaining alignment with regulatory expectations.

Within a release-by-exception framework, analytics function as enablers of structured decision making rather than as autonomous decision engines. Their primary value lies in making deviations visible, integrating diverse data streams, and supporting consistent application of predefined criteria within the PQS. By drawing together final QC results with manufacturing and in-process information, analytics help identify when a batch no longer fits within routine release pathways and must be evaluated through an exception process governed by written procedures and acceptance criteria.

One of the most important contributions of analytics is the early detection of deviations. Trends in process performance and product quality monitoring outputs can reveal variability that is not apparent from a single test result, allowing quality organizations to recognize exceptions based on patterns rather than isolated data points. This supports a more disciplined and reproducible trigger mechanism for escalation into risk-based evaluation pathways, consistent with quality system expectations for ongoing monitoring and life cycle control.

Analytics also support data integration across QC and manufacturing functions. Release decisions increasingly depend on multiple categories of evidence, including final test results, in-process measurements, and historical process knowledge. By providing a structured way to assemble and review this information, analytics strengthen the consistency of exception evaluations and reduce reliance on informal or fragmented data reviews. When used appropriately, PAT and real-time data concepts can contribute to this integrated evidence base while remaining embedded within the same regulatory framework that governs conventional testing and release.^1,6

Equally important is what analytics should not replace. Defined specifications and acceptance criteria remain the foundation of batch release and cannot be substituted by predictive models or dashboards.¹ Quality assurance authority over release decisions must also be preserved, as regulatory accountability for product quality resides within the quality system rather than in analytical tools or automated workflows. Risk-based judgment, grounded in quality risk management principles, remains central to evaluating exceptions and determining whether available evidence is sufficient to justify release.

Analytics therefore operate as decision support mechanisms, not decision makers. They enhance documentation and traceability by capturing how data were interpreted and how conclusions were reached within the PQS, but they do not absolve organizations of regulatory responsibility for those conclusions. In a release-by-exception model, the role of analytics is to strengthen the rigor and transparency of quality decisions while leaving ultimate authority with the governed quality system.

The central implication is that analytics improve the quality of release decisions without redefining their ownership. They provide structure, consistency, and visibility, but they do not alter the regulatory requirement that batch release be grounded in predefined specifications, risk-based evaluation, and accountable quality governance. In this sense, analytics enhance decision quality; they do not substitute for regulatory responsibility.

Translating a release-by-exception framework from concept into practice requires deliberate design of workflows, data systems, and organizational roles. For both sponsors and contract development and manufacturing organizations (CDMOs), the goal is not to create a new layer of complexity but to embed structured decision making within existing quality system expectations for testing, documentation, and risk management.

An effective release-by-exception model begins with a clearly defined decision workflow. Ownership of routine release and exception evaluation must be explicit, with responsibilities distributed across quality assurance (QA), QC, and manufacturing in a way that preserves accountability within the PQS. Escalation pathways should be defined in advance so that exceptions are handled through formal processes rather than ad hoc discussions. Pre-approved decision criteria anchored in established specifications and written procedures help ensure that similar situations are evaluated consistently over time, reducing variability in judgment and reinforcing regulatory defensibility.

A release-by-exception framework depends on the ability to assemble and interpret multiple categories of data in a reliable way. Data integrity is foundational, as release decisions rely on accurate and complete information drawn from testing and manufacturing records. Interoperability between systems used by QC and manufacturing is equally important, since exception evaluation often requires integrating final test results with in-process and historical performance data. Traceability from process observations to release decisions supports both internal governance and external inspection readiness by demonstrating how conclusions were reached within the PQS.

As products and processes evolve, acceptance criteria and decision thresholds must also be reassessed. A release-by-exception framework must therefore be compatible with change management and comparability strategies. Updating acceptance criteria over time requires documented justification and alignment with product knowledge gained across development and manufacturing. Process changes introduce new sources of variability that must be evaluated within the same risk-based decision structure used for routine release. Maintaining regulatory alignment under these conditions depends on ensuring that decision logic remains anchored in predefined specifications and quality risk management principles rather than shifting informally with operational pressures.

Finally, successful implementation depends on the capabilities of the people responsible for applying the framework. Exception evaluation requires more than technical test execution; it demands skills in data interpretation, understanding of process behavior, and familiarity with risk-based quality principles. Statistical and data literacy become increasingly important as analytics are incorporated into release decisions, and cross-functional collaboration between QA, QC, and manufacturing must be cultivated to support consistent judgment. In this sense, release-by-exception is as much an organizational discipline as it is a technical one, reinforcing the role of the PQS as the central authority for quality-related decisions.

Analytics-driven release is not achieved through tools alone. It depends on coherent workflows, reliable data infrastructure, disciplined change management, and a workforce equipped to apply structured, risk-based reasoning within established quality governance systems.

Relationship to PAT and Real-Time Release

Real-time release testing (RTRT) and PAT are defined within existing regulatory frameworks as approaches that use process data and measured material attributes to support evaluation of product quality. Regulatory guidance makes clear that real-time release, as described within the PAT framework, can meet the requirements for testing and release for distribution when implemented with prior regulatory approval and within established quality system controls.^1,6 In this sense, RTRT is not a separate regulatory pathway but an alternative means of generating the evidence required for release decisions under the same statutory expectations that govern traditional testing.

Within a release-by-exception framework, RTR and PAT function as supporting elements rather than as primary decision drivers. Process data and in-line or at-line measurements may contribute to the authorized data inputs used to assess whether a batch conforms to predefined specifications. When results fall outside expected ranges or display unexpected variability, these data streams can inform structured exception evaluation alongside final QC testing and historical process knowledge.^5,6 RTR therefore fits naturally within an exception-based model as an additional layer of evidence rather than as a replacement for acceptance criteria or quality governance.

For autologous cell therapies, practical constraints reinforce the need for hybrid approaches. Sterility and potency requirements remain central to release decisions for biological products, and these tests continue to rely on established methods that are not fully amenable to real-time substitution.^2,3 As a result, process data may support release justification, but they do so in conjunction with conventional quality control results and risk-based evaluation rather than through fully automated real-time approval. This positioning preserves alignment with PQS expectations while recognizing the operational realities of patient-specific products.

Looking forward, the role of process data in release justification is likely to expand as product knowledge and monitoring systems mature. Regulatory guidance already recognizes that quality can be evaluated using combinations of process controls and measured attributes when appropriately validated and governed.^6,8 In a release-by-exception model, this evolution does not eliminate the need for structured decision making but instead strengthens it by increasing the amount and quality of information available for exception evaluation.

Seen in this light, PAT and RTR do not redefine batch release for autologous cell therapies. They extend the evidence base used within established quality systems and reinforce the central premise of release-by-exception: that disciplined, risk-based decision frameworks, rather than isolated test results or automated outputs, remain the foundation of regulatory-aligned release decisions.

A release-by-exception framework offers a way to strengthen batch release without redefining regulatory responsibilities or introducing unproven automation. Its primary benefit is improved consistency in how release decisions are made when results fall outside routine expectations. By relying on predefined acceptance criteria, authorized data inputs, and structured escalation pathways, the framework reduces variability in judgment across batches and programs. Decisions are guided by documented procedures rather than informal interpretation, aligning operational practice with regulatory expectations for written sampling, testing, and acceptance criteria.

Another key advantage is stronger documentation and traceability. Exception evaluations are conducted within the PQS, with the rationale for each decision captured and retained as part of the quality record. This creates a transparent link between observed data, risk-based evaluation, and final release outcomes. Such traceability supports inspection readiness and reinforces the role of the PQS as the governing authority for quality-related decisions

The framework also enables better use of manufacturing and in-process data. Rather than treating final QC results in isolation, release-by-exception incorporates broader process performance and product quality monitoring outputs into structured evaluations. This integrated view allows organizations to contextualize deviations using historical process knowledge and measured trends, strengthening the scientific basis for release decisions without departing from established regulatory structures.

Over time, these practices contribute to increased regulatory confidence. A system that consistently applies predefined criteria, documents exception handling, and demonstrates risk-based decision making provides regulators with clearer evidence that product quality is being managed through disciplined governance rather than ad hoc judgment. This confidence is built not through claims of automation, but through visible alignment with quality system principles and life cycle monitoring expectations.

Finally, a release-by-exception framework supports life cycle learning and continuous improvement. Trending exception events and reviewing their underlying causes creates feedback loops that inform process refinement and potential adjustment of acceptance criteria as product knowledge grows. In this way, batch release evolves from a series of isolated decisions into part of an ongoing learning system that strengthens both manufacturing control and quality oversight across the product life cycle.

Batch release for autologous cell therapies is moving beyond a narrow focus on test results toward a model centered on structured decision making within the PQS. Regulatory expectations have long required that release be grounded in predefined specifications, written procedures, and acceptance criteria supported by appropriate statistical quality control principles. What is changing is not the regulatory foundation but the way organizations integrate growing volumes of manufacturing and quality data into governed release decisions.

Autologous cell therapies make this shift unavoidable. Patient-specific production, inherent variability, and tight clinical timelines expose the limits of purely test-based approval models, particularly when sterility and potency constraints introduce uncertainty that cannot always be resolved through additional testing alone. These realities demand structured exception handling that applies quality risk management principles rather than ad hoc judgment when results fall outside routine pathways.

Within this context, analytics do not replace quality systems; they operate in service of them. Process data, monitoring outputs, and historical performance trends strengthen the evidence base used for exception evaluation, but authority over release decisions remains anchored in defined specifications and quality governance. This positioning preserves regulatory accountability while enabling more consistent and transparent decision making.

A release-by-exception framework provides a scalable and defensible path forward. By embedding predefined criteria, risk-based evaluation, and documented rationale within the PQS, it transforms batch release from a single checkpoint into a disciplined decision framework that supports life cycle learning and continuous improvement. The future of batch release therefore lies not in automation alone, but in decision science: frameworks that integrate data, risk, and regulatory accountability to manage uncertainty in a controlled and repeatable way.

1. 21 CFR 211.165 — Testing and release for distribution. Code of Federal Regulations. Accessed 26 Jan 2026. https://www.ecfr.gov/current/title-21/chapter-I/subchapter-C/part-211/subpart-I/section-211.165

2. 21 CFR 610.12 — Sterility. Code of Federal Regulations. Accessed 26 Jan 2026.

3. “Potency Tests for Cellular and Gene Therapy Products: Final Guidance for Industry.” U.S. Food and Drug Administration. Jan. 2011.

4. ICH Q9 — Quality Risk Management. International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use. 1 Mar. 2025.

5. ICH Q10 — Pharmaceutical Quality System. International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use.

6. Guidance for Industry: PAT — A Framework for Innovative Pharmaceutical Development, Manufacturing, and Quality Assurance. U.S. Department of Health and Human Services. Sep. 2004.

7. “Sterility Tests.” U.S. Pharmacopeia. Accessed 26 Jan. 2026.

8. Guideline on Real Time Release Testing. European Medicines Agency. 29 Mar. 2012.