跳到正文
BioProcess International · 工艺与质量· Naveenganesh Muralidharan, Akshay Phulgirkar·AI 评分63

细胞与基因治疗产品降低体积生物负载检测的监管与风险统计方法

A Regulatory- and Risk-Based Statistical Approach to Reduced-Volume Bioburden Testing for Cell and Gene Therapy Products

AI 导读

作者提出以泊松统计和过滤突破概率指数为基础的评估框架,论证细胞与基因治疗产品在物料受限时可使用降低体积的重复取样开展生物负载检测。文章说明检出概率取决于总检测体积(单次体积乘以重复数)而非单一样品体积,在 0.1 CFU/mL 条件下三个 1 mL 重复样品的检出概率高于单个 1 mL 样品。作者认为降低体积的重复检测与经验证的除菌过滤结合,不增加患者风险,并符合无菌保证原则。

正文 · 原文

24-9-Muralidharan-P1.png

https://stock.adobe.com

The rapid growth of cell and gene therapy (CGT) has transformed the biopharmaceutical landscape, introducing products that are often patient-specific, produced at small scales, and associated with exceptionally high material value. Unlike traditional monoclonal antibodies (mAbs) or recombinant proteins that are manufactured at hundreds or thousands of liters, many CGT processes operate at volumes where every milliliter of product has tangible clinical and economic importance (1, 2).

Conventional microbiological control strategies, including in-process bioburden testing and compendial sterility testing, were designed for high-volume biologics manufacturing. That paradigm implicitly assumes that sampling volumes of 10–100 mL or more represent a negligible fraction of a given batch. In low-volume CGT contexts, however, such sampling can remove a meaningful proportion of available material, potentially reducing patient doses or increasing manufacturing cost.

Related:ANOVA Power Analysis Using Noncentral F-Distribution

Recent industry discussions have highlighted this mismatch between legacy testing expectations and modern CGT realities. Reports in trade literature describe cases in which microbiological testing volumes approach or exceed several percent of batch volume. Such findings are prompting reconsideration of traditional approaches and exploration of reduced-volume strategies that maintain microbial-risk control while conserving product.

This article reviews the statistical and regulatory basis for reduced-volume bioburden testing in CGT manufacturing. We focus on the probability of detecting microorganisms in small-volume samples using Poisson statistics, and we examine how sample volume and replicate testing can improve detection confidence while minimizing product loss. The goal is to provide an objective, literature-supported framework to inform risk-based decision-making for microbiological monitoring in material-constrained processes.

Statistical Basis of Presence/Absence Testing

When microorganisms are randomly distributed within a well-mixed liquid, their occurrence in any given test portion follows Poisson statistics. Under Poisson assumptions, the probability (P0) that a sample of volume v contains no microorganisms is given by Equation 1, in which c is the true bioburden concentration in colony-forming units per milliliter (CFU/mL) and v is the test volume (in mL):

P0 = e–(cv)

Equation 2 gives the probability of detecting at least one microorganism (PD≥1) in testing when n number of independent replicate samples are tested:

Related:Vendor Governance Is a Quality System: Moving Beyond Status-Based Oversight in Outsourcing and Technology Transfer

PD≥1 = 1 – e–(cvn)

Those relationships form the mathematical basis of presence/absence testing and underlie the classical Most Probable Number methodology described by Cochran (3). A key implication of this methodology is that detection probability depends on the product of sample volume and number of replicates — the total tested volume — rather than on individual sample size alone. That distinction is particularly relevant in low-volume contexts, in which increasing the size of a single sample might be impractical but dividing the total tested volume into multiple replicates might be feasible.

Representative Bioburden Regimes for Risk-Based Evaluation

In aseptically processed systems, acceptable bioburden levels vary across the manufacturing continuum. Early steps may tolerate relatively higher levels due to subsequent clearance mechanisms, whereas later steps approaching drug-substance and drug-product stages require increasingly stringent control.

To contextualize sampling performance, we consider three representative regimes that are commonly discussed in the literature (4, 5):

  • 1 CFU/mL, reflecting readily detectable contamination typical of early process steps

  • 0.1 CFU/mL (1 CFU/10 mL), representative of well-controlled aseptic intermediates

  • 0.01 CFU/mL (1 CFU/100 mL), approaching near-sterile conditions typical of late-stage operations.

Related:Sterilization as a Design Decision for Chemistry, Manufacturing, and Controls: Why Method Selection Must Be Integrated Early in Drug and Device Development

Those regimes are not regulatory limits, but practical reference points used to evaluate detection behavior across realistic microbial-control conditions.

Effect of Sample Volume and Replication on Detection Probability

Application of the Poisson model demonstrates how detection probability varies with both sample volume and replication (Figure 1).

At near-sterile conditions (0.01 CFU/mL), even large total tested volumes provide limited detection confidence, reflecting the inherent difficulty of detecting rare contamination events. No practical sampling strategy can guarantee detection at such low bioburden levels.

At intermediate conditions (0.1 CFU/mL), replicate sampling provides substantial benefit. For example, three 1-mL samples (total volume = 3 mL) yield markedly higher detection probability than a single 1-mL sample would, thereby approaching the performance of much larger single-volume tests.

At higher bioburden levels (1 CFU/mL), detection probability rapidly approaches unity even at small volumes, and incremental gains from additional volume or replication diminish.

Collectively, those behaviors demonstrate that

  • detection confidence is governed by total tested volume (v × n)

  • replication is particularly effective under realistic low-bioburden conditions

  • probabilistic limitations dominate as true bioburden approaches near-sterile levels.

Those trends align with published discussions supporting replicate-based strategies for conserving material while maintaining defined detection confidence.

24-9-Muralidharan-F1.png

Figure 1: Replication-driven improvement in microbial-detection probability across a continuum of bioburden conditions (CFU = colony-forming unit, PD≥1 = probability of detecting at least one microorganism in n replicates)

Integrated Probability Framework for Microbial Control

Equation 2 can be rewritten as Equation 3 to understand the probability of not detecting any microorganism or of generating false-negative results:

PND = 1 – PD≥1

The probability that microorganisms present in the presterile bulk challenge and breach the sterilizing filter can be expressed conservatively using the filter-breach probability index (FBPI), as shown in Equation 4:

24-9-Muralidharan-E4.png

where Vb is the presterile batch volume (in mL), c is the risk-based assumed bioburden concentration (in CFU/mL), and A is the effective filter surface area (in cm2). The denominator reflects typical validation challenges of 10⁷ CFU/cm2 for sterilizing-grade filters (4, 5).

Within this layered control framework, the overall probability of microbial escape (Pescape) can be estimated as shown in Equation 5:

Pescape = PND × FBPI

This formulation highlights that sterility risk is governed jointly by sampling-based nondetection and filter performance. When replicate-based sampling achieves high detection confidence (e.g., PND ≤ 0.05) and FBPI remains orders of magnitude below 10–6, the combined probability of microbial escape becomes vanishingly small and consistent with accepted sterility-assurance level (SAL) expectations.

Regulatory Perspectives on Risk-Based Microbial Control

Patient risk arises only when two independent protective layers fail concurrently: a false-negative bioburden test result and microbial passage through a validated sterilizing-filtration step. Sterility assurance is therefore governed by the combined probability of nondetection and filter breach rather than by either element in isolation.

When replicate-based reduced-volume bioburden testing achieves a defined nondetection probability and is integrated with sterilizing-grade filtration that is validated to retain ≥107 CFU/cm2, the resulting combined escape probability remains orders of magnitude below accepted SAL expectations. This integrated probabilistic framework demonstrates that reduced-volume bioburden testing, when coupled with validated filtration, does not increase patient risk and remains consistent with regulatory sterility-assurance principles.

24-9-Muralidharan-F2.png

Figure 2: Patient risk exists only at the intersection of independent control failures. (CFU = colony-forming unit, Pescape = probability of microbial escape, PND = probability of nondetection, SAL = sterility-assurance level)

A Practical and Defensible Approach

CGT manufacturing challenges traditional microbiological testing paradigms by introducing low-volume, high-value processes in which conventional sampling volumes can materially diminish product yield. Probabilistic detection theory demonstrates that presence/absence testing is governed by Poisson statistics and that replicate-based sampling achieves defined detection confidence based on total tested volume rather than individual sample size.

When integrated with conservative estimates of batch bioburden and validated sterile-filtration performance, reduced-volume replicate strategies can be contextualized within overall sterility-assurance objectives. Published literature and regulatory perspectives support risk-based, phase-appropriate adaptation of microbiological-control strategies for advanced therapies.

Reduced-volume bioburden testing, when scientifically justified and embedded within a comprehensive microbial control framework, represents a practical and defensible approach aligned with modern CGT manufacturing realities and risk-based pharmaceutical quality principles.

References

1 Eckford C. Reducing Bioburden Testing and Sterility Testing Volumes for Gene Therapies. Eur. Pharm. Rev. 23 June 2023; https://www.europeanpharmaceuticalreview.com/reducing-bioburden-testing-and-sterility-testing-volumes-for-gene-therapies/183816.article.

2 BioPhorum. Strategies for Minimizing the Impact of Bioburden and Sterility Testing on Gene Therapy Batch Yield. Cell & Gene 2 January 2024; https://www.cellandgene.com/doc/strategies-for-minimizing-the-impact-of-bioburden-and-sterility-testing-on-gene-therapy-batch-yield-0001.

3 Cochran WG. Estimation of Bacterial Densities by Means of the “Most Probable Number.” Biometrics 6(2) 1950: 105–116; https://doi.org/10.2307/3001491.

4 Bechtold-Peters K, et al. Risk-Based Approach to Setting Sterile Filtration Microbial Bioburden Limits — Focus on Biotech-Derived Products. Eur. J. Pharm. Biopharm. 198, 2024: 114151; https://doi.org/10.1016/j.ejpb.2023.11.016.

5 Muralidharan N, et al. Microbial-Control Hold-Time Assessments: Guidance on Matrix and Bracketing Validation Approaches. BioProcess Int. 23(3) 2025: 23–34; https://www.bioprocessintl.com/validation/microbial-control-hold-time-assessments-guidance-on-matrix-and-bracketing-validation-approaches.

BPI editorial advisor Naveenganesh Muralidharan is founder and principal consultant at Bench2Batch Life Cycle Partners LLC; info@Bench2Batchcom; https://www.bench2batch.com. Akshay Phulgirkar is senior manager of manufacturing sciences and technology (MSAT) at Moderna; https://moderna.com.

Please cite this article as Muralidharan M, Phulgirkar A. A Regulatory- and Risk-Based Statistical Approach to Reduced-Volume Bioburden Testing for Cell and Gene Therapy Products. BioProcess Int. 24(9) 2026: 240910.

来源:BioProcess International · 工艺与质量 · bioprocessintl.com