XGene CMC IntelligenceXGene Intelligence

Cell Therapy Quality Overall Summary — Writing Module 2.3 for an Incompletely Characterized Drug

Starting MaterialsSpecificationsBiologicsGene TherapyCell Therapy

The QOS for a cell therapy product must summarize a drug that is, by definition, incompletely characterized. The way you handle that reality in Module 2.3 determines whether CBER trusts…

By Khaled Aamer, PhD · Founder, XGene LLC Aug 22, 2026 10 min read
On this pageArticle overview

    ─────────────────────────────────────────────────────────────────────────────

    The QOS for a cell therapy product must summarize a drug that is, by definition, incompletely characterized. The way you handle that reality in Module 2.3 determines whether CBER trusts your entire CMC package.

    That single tension — the inherent biological variability of a living therapeutic versus the regulatory expectation of a complete, coherent quality summary — defines every cell therapy CMC submission. The Quality Overall Summary is not a table of contents for Module 3. It is the document that tells CBER’s Office of Therapeutic Products (OTP, formerly the Office of Tissues and Advanced Therapies) whether the applicant truly understands the product, including what cannot be pinned down with a number. When that narrative is missing or evasive, the result is not a minor deficiency — it is a loss of reviewer confidence that propagates across every subsequent module review.

    ─────────────────────────────────────────────── The CT Module 2.3 Architecture: What the QOS Must Accomplish for a Product That Cannot Be Fully Characterized ───────────────────────────────────────────────

    ICH M4Q(R1) defines the Quality Overall Summary as a critical synopsis of the information in Module 3, written to allow a reviewer to understand the quality of the product without turning to the full data sections unless specific detail is required. For a conventional small molecule, that synopsis is achievable because the drug substance is structurally defined, synthesis is reproducible, and characterization is essentially complete. For a cell therapy product — whether an autologous CAR-T manufactured from a patient’s leukapheresis starting material, an NK-92-based allogeneic off-the-shelf platform, or an iPSC-derived NK cell with a 35–42-day differentiation timeline — the product is a living biological system whose full characterization is neither achievable nor expected at time of approval.

    The Module 2.3 architecture for a cell therapy product must therefore accomplish something more sophisticated than summarizing data. It must frame the product’s biological complexity for the reviewer, explain which quality attributes are well-characterized versus incompletely understood, and demonstrate that the manufacturing and control strategy is designed to manage that uncertainty rather than obscure it. ICH Q8(R2) provides the conceptual foundation for this approach through its enhanced pharmaceutical development framework: the applicant is expected to identify critical quality attributes, understand the relationship between process parameters and product quality, and design controls that provide assurance within the acknowledged limits of characterization. Applying Q8(R2) principles explicitly in the QOS narrative — not just in Module 3’s pharmaceutical development section — signals to the CBER reviewer that the applicant has internalized the framework rather than using it as a documentation checkbox.

    The most common architectural failure is a QOS written as a section-by-section abstract of Module 3 without an integrating narrative. A reviewer reading such a document receives information but no argument. The argument the QOS must make is this: the product has inherent biological variability, that variability is bounded by the specification system and manufacturing controls, the control strategy is risk-proportionate, and the totality of the CMC package provides reasonable assurance of consistent quality. Without that argument explicitly constructed, the QOS fails its primary regulatory function regardless of how well-organized the Module 3 data sections are.

    ─────────────────────────────────────────────── Manufacturing Summary and Potency Data Overview: The CT QOS Sections CBER Reviews Most Carefully ───────────────────────────────────────────────

    CBER’s review of the manufacturing summary section focuses on whether the QOS accurately represents the critical steps and controls described in Module 3, and whether the controls are appropriate for the specific cell therapy platform. For an NK-92-based allogeneic product, the manufacturing summary must address not only expansion in suspension bioreactor with IL-2 at 200–500 IU/mL and CAR expression verified by flow cytometry and VCN confirmed by ddPCR, but also the irradiation step — administered at 1,000–2,000 cGy confirmed by dosimeter — that eliminates replication capacity before patient infusion. The QOS must explain why this step is critical: NK-92 is an established NK cell line derived originally from a non-Hodgkin’s lymphoma patient and remains IL-2 dependent and proliferatively competent, making residual replication capacity a safety-critical quality attribute. The irradiation validation — including colony-forming assay post-irradiation with an acceptance criterion of no colonies — must be summarized in the manufacturing section of Module 2.3 with enough mechanistic context that the reviewer understands why the irradiation confirmation test appears on the lot release panel.

    The potency section of the QOS is where CBER deficiencies are most concentrated, and where the failure pattern is most predictable. Potency for a cell therapy product must be linked to mechanism of action under ICH Q11 and FDA’s Chemistry, Manufacturing, and Controls Information for Human Gene Therapy INDs (2020) — a guidance that, while directed at gene therapy, is routinely applied by CBER reviewers to cell therapy products with genetic modifications such as CAR-NK constructs. The operational failure scenario that produces a CBER information request is not subtle: a QOS that states “a potency assay is under development” in the potency section without specifying the functional assay to be qualified, the acceptance criterion under consideration, and a milestone timeline for assay validation. For an NK-92 CAR-T targeting CD19, the cytotoxicity assay against a CD19-positive target cell line with an acceptance criterion of ≥30% killing at a defined effector-to-target ratio — consistent with the K562-based cytotoxicity benchmark used for lot release at E:T 5:1 — represents the kind of mechanism-of-action-linked functional readout CBER expects to see described and justified in Module 2.3, not deferred without a plan.

    The QOS potency narrative must do more than name the assay. It must explain how the assay result connects to the proposed clinical mechanism, why the acceptance criterion is clinically meaningful, and what happens to a lot that fails the criterion. A potency section that presents the acceptance criterion as a number without any of that narrative context invites the reviewer to question whether the applicant understands the product’s biology — which is precisely the loss of reviewer confidence that triggers a full CMC review cycle.

    ─────────────────────────────────────────────── Comparability and Consistency Data in the CT QOS: How CBER Assesses Lot-to-Lot Quality ───────────────────────────────────────────────

    Autologous cell therapy products present a QOS challenge that has no equivalent in any other pharmaceutical category: every lot is manufactured from a different patient’s cells, meaning lot-to-lot variability is not a manufacturing deviation — it is an inherent product characteristic. EMA’s Guideline on Human Cell-Based Medicinal Products (EMEA/CHMP/410869/2006) addresses this directly by requiring that the specification system be designed to accommodate the expected range of biological variability while still providing meaningful quality assurance. In the QOS, this means the comparability and consistency section cannot be written as if the product were a monoclonal antibody. It must explicitly describe the expected distribution of quality attribute values across clinical lots, explain which attributes are patient-dependent versus process-controlled, and demonstrate that the in-process controls and release testing panel are positioned to catch lots that fall outside the acceptable range — not merely to confirm that every lot looks identical.

    For allogeneic platforms — cord blood-derived NK cells, NK-92 cell lines, or iPSC-derived NK cells with their 35–42-day differentiation timeline — lot-to-lot consistency is a more tractable claim, but the QOS must still address the MCB/WCB paradigm under ICH Q5D, including the extensive characterization required at each banking tier. For CB-NK products, the QOS consistency section must acknowledge that CD56bright percentage and HLA-mismatched tolerance are donor-dependent variables, and explain how the manufacturing process controls — including CD3/CD56 enrichment and the cytokine expansion cocktail — are designed to produce a consistent functional profile despite donor variability. The CD56+CD3− identity criterion of ≥90% at lot release, supported by flow cytometry, represents a specification anchor for the consistency narrative that CBER expects to see justified rather than simply stated.

    The consistency section of the QOS is where the integration between Module 2.3 and Module 3’s comparability protocols must be explicitly demonstrated. FDA’s Guidance for Industry: CMC Postapproval Manufacturing Changes for Specified Biological Products To Be Documented in Annual Reports (2021), applying the 21 CFR 601.12 annual reporting framework, reinforces that CBER tracks product consistency across the clinical development timeline and expects the QOS to evolve as additional lot data accumulates. A QOS written at IND phase should include a commitment to expanding the consistency dataset; a BLA QOS should present the complete lot history with an integrated assessment of attribute trends. When the QOS omits this longitudinal perspective, the reviewer cannot assess whether the process is in a state of control — which is the fundamental question the comparability section must answer.

    ─────────────────────────────────────────────── [FRAMEWORK BOX]

    Writing a Cell Therapy QOS That Guides the CBER Reviewer Through a Complex BLA Module 3 ───────────────────────────────────────────────

    The XGene Cell Therapy QOS Narrative Architecture is a section-by-section writing framework for cell therapy INDs and BLAs that structures the required narrative elements for each CMC domain, provides guidance on handling inherent biological variability, and includes internal consistency validation checks that prevent CBER information requests.

    1. Characterization Boundary Mapping. Before drafting any section of the QOS, identify and document which critical quality attributes are fully characterized, which are partially characterized with ongoing work, and which are acknowledged as incompletely characterized with a defined monitoring plan. This map becomes the organizing logic of the QOS narrative — it tells the reviewer what the applicant knows, what is still being learned, and why the current state of knowledge is sufficient to proceed.

    2. Mechanism-Linked Potency Section Construction. Draft the potency section by starting with the proposed clinical mechanism of action, identifying the specific functional assay that measures that mechanism, stating the acceptance criterion with its scientific rationale, and connecting the criterion explicitly to the clinical dose and patient population. For a CAR-NK-92 construct targeting CD19, this means linking the cytotoxicity acceptance criterion at E:T 5:1 to the expected in vivo effector-to-target ratio at the proposed cell dose — not presenting the criterion as a standalone number.

    3. Lot-to-Lot Variability Narrative Integration. For autologous products, write the consistency section to distinguish patient-dependent variability from process-dependent variability, specify which attributes fall into each category, and explain how the specification system accommodates patient-dependent variability while still rejecting out-of-specification lots on process-dependent attributes. For allogeneic products, present the MCB/WCB characterization data as the consistency anchor and explain how donor variability at early manufacturing steps is controlled by downstream in-process specifications.

    4. Cross-Reference Verification Against Module 3. Before finalizing any QOS section, validate every cross-reference against the actual Module 3 section number, table number, and page range in the current version of the eCTD. Inconsistent cross-references are among the most common sources of CBER information requests — they signal to reviewers that the Module 2.3 was written independently of Module 3 rather than as an integrated summary, which undermines confidence in the entire CMC package.

    The output of the XGene Cell Therapy QOS Narrative Architecture is a complete Module 2.3 draft with annotated narrative rationale for each section — a document that functions as both the regulatory submission text and an internal reference that the CMC team can use to respond to CBER information requests without reconstructing the argument from scratch.

    ───────────────────────────────────────────────

    A cell therapy CMC package that reaches BLA review with an inadequate Module 2.3 does not simply generate information requests — it trains the reviewing division to distrust the applicant’s characterization of their own product. Once that doubt is established, it is expensive and slow to reverse: every subsequent response must work against the prior impression created by the QOS. The cost of a CBER complete response letter driven by CMC deficiencies in a cell therapy BLA is not measured only in review cycle delays; it is measured in manufacturing hold time, clinical site uncertainty, and the compounding difficulty of negotiating specification revisions under post-review scrutiny. The QOS is the only document in the eCTD that gives the applicant the opportunity to guide the reviewer’s interpretation of the entire Module 3 before the data is examined — writing it as a passive summary rather than an active argument is one of the most consequential decisions a cell therapy CMC team makes.

    In your current cell therapy IND or BLA Module 2.3, can you identify the section that explicitly addresses expected lot-to-lot variability for your autologous product and explains how the specification system accommodates that variability — and whether the potency section includes a mechanism-of-action narrative linked to the specific functional assay used for lot release?

    Primary regulatory references