Cell Therapy CMC — Why the Module 3 Architecture for SM Will Fail Your CBER Reviewer

Cell therapy CMC packages fail at CBER not because the science is wrong, but because the regulatory architecture is borrowed from the wrong framework.

When a small molecule or even a recombinant protein program encounters a Module 3 CMC deficiency, the path to resolution is usually a matter of generating additional data or clarifying a manufacturing step. For a cell therapy program, an architectural deficiency — a fundamental mismatch between how the submission is organized and how CBER’s Office of Tissues and Gene Therapies (Office of Therapeutic Products (OTP)) actually reviews it — can trigger a clinical hold that no additional data will resolve quickly. The consequence is not a request for information; it is a suspended IND and a program that has lost months of development time before a single patient is dosed.

Why the Module 3 Architecture That Works for Small Molecules Fails CBER’s Cell Therapy Reviewers

The standard eCTD Module 3 architecture was designed around a paradigm of defined chemical entities or well-characterized biological molecules that can be isolated from their manufacturing process and tested as independent substances. Sections 3.2.S and 3.2.P map cleanly onto a drug substance synthesized in a reactor, purified to specification, and transferred to a drug product formulation step. Cell therapy products do not fit this paradigm, and the failure to recognize this at the architecture stage — not the data stage — is the most consequential CMC error made by programs entering CBER review for the first time.

The most common and operationally damaging version of this failure occurs when a CAR-T program defines the drug substance as the lentiviral vector used to transduce T cells, rather than the transduced cell product itself. This is not a semantic error. A reviewer at CBER Office of Therapeutic Products (OTP) reading a 3.2.S section devoted to vector production and a 3.2.P section devoted to the formulated cell bag is looking at a submission that has effectively split the drug across two sections — placing the manufacturing critical quality attributes in neither location coherently. FDA’s Guidance for Industry: Considerations for the Development of Chimeric Antigen Receptor (CAR) T Cell Products (CBER, finalized January 2024) is explicit that the cellular product itself is the biological product subject to licensure, not the manufacturing inputs, however complex. When this is wrong in the architecture, every subsequent section — specifications, analytical procedures, stability — is built on a misidentified substrate.

The second structural failure is the absence of a controlled starting material section for the autologous apheresis product. For a CAR-T program using leukapheresis as patient-derived starting material, a reviewer expects to find the apheresis product characterized with defined acceptance criteria within the CMC package — not assumed to be a clinical variable outside CMC scope. A leukapheresis product collected with a target of at least 5×109 total nucleated cells (TNC) and at least 1×109 CD3+ T cells with viability at or above 80% represents a defined starting specification. Omitting this section, or subordinating it to a clinical procedure note, leaves the reviewer with no CMC basis on which to assess input quality — and that gap propagates forward into every downstream process performance claim.

The Living Drug Problem: How the Nature of Cell Therapy Products Changes Every CMC Expectation

A small molecule drug substance is inert between manufacturing and administration. You can test it, hold it, retest it, and its identity does not change. A leukapheresis product collected from a patient with relapsed/refractory lymphoma and shipped fresh at 2–8°C for no more than 24 hours before arriving at the manufacturing site is a living, dynamic starting material whose composition is already shifting at the moment of collection. Cryopreservation at −196°C in liquid nitrogen vapor phase can extend the handling window, but it introduces a 10–20% cell loss and measurable T cell phenotype changes — including increased differentiation — that alter the manufacturing input. Neither of these handling scenarios has an analogue in small molecule or recombinant protein CMC, and neither can be addressed by repurposing a standard raw material specification.

This is what regulators mean when they describe cell therapy manufacturing as process-as-product. The ICH Q6B guideline, which governs specification-setting for biotechnological products, provides the conceptual framework for setting specifications based on manufacturing process knowledge in the absence of a fully characterized reference standard. For cell therapy, this framework must be extended further — because the starting material (the patient’s own cells) is itself variable in ways that no manufacturer fully controls. The specification system for a CAR-T product, including parameters such as target CAR expression of 20–60% by flow cytometry, vector copy number at or below 5 by ddPCR, and a final product yield of at least 5×108 CAR+ T cells per patient dose, must be set prospectively, before the clinical data exists to justify it empirically.

The FDA CMC Information for Human Gene Therapy INDs Guidance (2020) extended its scope to include gene-modified cell therapy products and addresses this challenge explicitly — acknowledging that early-phase specifications are necessarily based on manufacturing capability rather than clinical outcome correlation. What it does not do, and what many programs misread, is provide a format template for how the CMC package should be organized to address cell therapy-specific controls. That organizational architecture must be designed by the submitting team, and it must anticipate the reviewer’s logic — which is categorically different from the logic applied to a chemically synthesized product.

The Four Module 3 Sections Where Cell Therapy CMC Is Fundamentally Different From All Other Modalities

The four sections where cell therapy CMC diverges most sharply from any other modality — and where CBER deficiency letters are most frequently generated — are drug substance definition, starting material control, specifications, and the donor eligibility and infectious disease testing framework. In the drug substance definition section, the transduced T cell product must be identified as the drug substance, with the lentiviral vector addressed as a manufacturing reagent with its own defined quality controls — not as the drug substance itself. This requires a deliberate architectural choice to create a dedicated section for vector characterization within the manufacturing controls framework, not within 3.2.S. The EMA/CAT Guideline on quality, non-clinical and clinical aspects of medicinal products containing genetically modified cells reinforces this same logic for ATMP submissions, where ATMP classification must be established before the submission is organized — because classification determines the regulatory framework under which every section is assessed.

In the specifications section, the requirements of ICH Q6B apply, but must be interpreted in the context of a living cellular product for which stability is measured in post-thaw viability recovery and functional potency assays, not in chemical degradation profiles. The activation step using Dynabeads Human T-Activator CD3/CD28 at a 3:1 bead-to-cell ratio over 24–48 hours at 37°C in 5% CO2 with IL-2 at 100–300 IU/mL, followed by lentiviral transduction at a multiplicity of infection of 3–10 with RetroNectin fibronectin fragment CH-296 at 10 μg/cm2, represents a process sequence in which every parameter is simultaneously a manufacturing control and a product quality determinant. The specification system must capture this linkage; a table of release tests without process parameter context is not a defensible specification in CBER’s framework.

The fourth and most commonly omitted section is the donor eligibility and infectious disease testing framework under 21 CFR Part 1271. For autologous cell therapy products, 21 CFR 1271.90(a)(2) exempts the donor from the formal donor-eligibility determination and screening/testing otherwise required under §§1271.50, 1271.75, 1271.80, and 1271.85 — but the product must still be labeled “For Autologous Use Only,” and FDA’s CAR T cell product guidance recommends infectious disease testing as a matter of manufacturing facility safety and cross-contamination control, independent of the 1271 exemption. CBER reviewers expect this distinction — what 1271 requires versus what is tested for other risk-management reasons — to be addressed explicitly in the CMC package, not deferred to clinical operations, because the apheresis collection and any infectious disease testing occur before the manufacturing process begins, and their adequacy is a precondition for the legitimacy of the manufactured product.

Building a CT CMC Strategy That CBER Will Engage With Productively From Phase 1

The XGene Cell Therapy CMC Architecture Blueprint is a Module 3 structural design framework for cell therapy products that maps each CMC section to the cell-specific regulatory requirements, drug substance and drug product definitional requirements, and CBER/EMA section-specific content expectations before the first page of the IND is drafted.

Step 1 — Drug Substance Boundary Setting: Confirm in writing, before any section is drafted, that the drug substance is the transduced cell product and that the viral vector is scoped as a manufacturing reagent with its own quality controls. This is not a labeling decision — it is an architectural decision that determines where critical quality attribute data lives in the submission, how the specification table is structured, and which section the reviewer reads first when evaluating product safety.

Step 2 — Starting Material Architecture: Build a dedicated starting material section for the apheresis or biopsy product that includes collection criteria, transport conditions, acceptance specifications (including minimum TNC and CD3+ cell count and viability thresholds), and the infectious disease testing panel performed as a risk-management measure — noting explicitly where the formal donor-eligibility determination is exempted for autologous use under 21 CFR 1271.90(a)(2). This section must be written as a CMC control section, not a clinical procedure.

Step 3 — Process-Linked Specification Design: Construct the specification table as a process-annotated document — each release test linked to the manufacturing step that most directly determines it, each acceptance criterion anchored to a manufacturing capability range established during process development. For parameters such as CAR expression by flow cytometry and vector copy number by ddPCR, the specification is not a regulatory compliance checkbox; it is the primary evidence that the manufacturing process performed within its validated operating range.

Step 4 — CBER/EMA Section Crosswalk: Produce a section-by-section crosswalk that maps each Module 3 section to the specific regulatory expectation under CBER Office of Therapeutic Products (OTP) review (for IND and BLA) and EMA CAT review (for CTA and MAA), identifying where the same data must be presented differently to satisfy the two agencies’ format and content requirements simultaneously.

The output of this framework is a pre-submission CMC architecture package — a structured map of every Module 3 section, its content requirements under CBER and EMA review, and the specific data elements that must be generated or confirmed before the submission is drafted — so that deficiencies are resolved before the submission is filed, not in response to a clinical hold letter.

The cost of architectural misalignment in a cell therapy CMC package is not measured in additional experiments; it is measured in clinical hold delays, lost manufacturing slots, and the compounding effect on a patient population that is often out of treatment options. A program that enters CBER review with a Module 3 architecture borrowed from a small molecule precedent will receive deficiency questions that cannot be answered by data alone — because the questions will concern the structure of the submission, not its content. Rebuilding a CMC package under a clinical hold, while maintaining GMP manufacturing, managing patient scheduling, and responding to information requests on an accelerated timeline, is a qualitatively different challenge from writing the package correctly at the outset. The programs that navigate CBER’s cell therapy review productively from Phase 1 are the ones that made architecture decisions deliberately, with a clear understanding of what CBER’s reviewer needs to find in each section and where they expect to find it.

In your current cell therapy IND CMC package, can you confirm today that the drug substance is defined as the cell product (not the vector used in manufacturing), that the starting material (apheresis or biopsy) is characterized and controlled in a dedicated section, and that donor eligibility and infectious disease testing requirements under 21 CFR 1271 are addressed?