Clinical trials for CAR-T cell therapies are described in terms of patient outcomes: response rates, duration of remission, cytokine release syndrome incidence, progression-free survival. These are the variables that appear in publications, investor presentations, and regulatory submissions. They are the endpoints that determine whether a program succeeds.
What they do not capture — what almost no external discussion of CAR-T development captures — is the frequency with which clinical outcomes are determined before the patient ever receives the product. The manufacturing failure rate in CAR-T development is not a minor footnote. It is a program-defining variable that shapes every aspect of clinical trial design, patient selection, site logistics, and ultimately commercial viability. And it is almost entirely a function of manufacturing control.
The Starting Material Problem
CAR-T manufacturing begins with a biological input that is irreducibly variable: the patient’s own cells. For autologous products — which include every currently approved CAR-T therapy — leukapheresis material collected from a heavily pre-treated oncology patient is not a standardized raw material in any conventional manufacturing sense. It is a biological sample whose quality, cell composition, viability, and expansion potential reflect the patient’s disease history, prior treatment regimens, immune status at the time of collection, and the conditions and timing of the collection procedure itself.
Patients who have received multiple prior lines of chemotherapy — the population that most CAR-T trials enroll — may have compromised T cell populations. Their cells may expand poorly in culture, transduce at lower efficiency, or produce a final product with a phenotype that predicts inferior in vivo persistence. Some patients’ cells simply fail to produce a manufacturable product. The FDA’s 2024 CAR-T guidance explicitly acknowledges this: it recommends establishing acceptance criteria for leukapheresis starting material — minimum cell number, viability, CD3+ percentage — specifically to identify starting material that is unlikely to support successful manufacturing before the process begins.
This recommendation has a direct clinical implication that is rarely made explicit in trial design discussions. If a patient’s leukapheresis material does not meet acceptance criteria, the manufacturing attempt either fails or produces a product that fails lot release. The patient receives no treatment. In a disease context where patients are typically in active progression during the manufacturing window — which for autologous CAR-T typically runs two to four weeks from leukapheresis to product release — a manufacturing failure is not a recoverable event for many patients. It is a clinical outcome.
The guidance also flags a specific complexity for patients who have received prior CAR-T therapy: their leukapheresis starting material may contain previously administered CAR-T cells, which can affect expansion, transduction efficiency, and final product potency in ways that differ from treatment-naive patients. Developers running trials that enroll both prior CAR-T recipients and treatment-naive patients are recommended to evaluate and account for these differences in study design and analysis — a requirement that adds analytical complexity to an already demanding population.
Lot-to-Lot Variability: The Consistency Challenge at Scale
Even when starting material meets acceptance criteria and manufacturing proceeds without failure, CAR-T products exhibit lot-to-lot variability that has no equivalent in conventional biologic manufacturing. Each lot is manufactured from a unique biological input, processed through a multi-step procedure involving cell selection, activation, transduction, expansion, formulation, and cryopreservation. Variability can be introduced at every step, and the cumulative effect on the final product — in terms of cell phenotype, CAR expression level, vector copy number, potency assay output, and in vivo expansion potential — can be substantial.
The FDA’s guidance frames lot-to-lot variability not as an inherent characteristic to be accepted but as a manufacturing control problem to be systematically addressed. Critical process parameters — the manufacturing variables that have the most significant impact on product quality — should be identified during process development and used to establish in-process controls. The guidance provides specific examples: fixed bead-to-cell ratios for T cell activation, constant vector amounts per cell for transduction, fixed electroporation settings for non-viral gene delivery. These are not arbitrary process constants. They are the variables whose control most directly determines product consistency.
In-process testing — measuring viability, cell number, phenotype, and CAR expression at defined points during manufacturing — allows manufacturers to make informed decisions during the process rather than discovering quality deviations at final product release. The guidance explicitly frames in-process testing as a manufacturing control tool, not merely a documentation exercise.
The commercial implication of uncontrolled lot-to-lot variability extends beyond regulatory risk. For autologous products, each patient’s product is a single lot. A product that is within specifications but at the lower end of the potency range produces different clinical outcomes than one at the upper end. If variability in product quality correlates with variability in clinical outcomes — and there is emerging evidence that T cell phenotype at the time of infusion does correlate with response durability — then manufacturing control is not just a regulatory issue. It is a clinical efficacy issue.
Manufacturing Failure Contingency Planning: The Requirement Nobody Budgets For
One of the most operationally significant — and consistently underbudgeted — requirements in the FDA’s CAR-T guidance is the recommendation to build a manufacturing failure contingency plan into the clinical protocol from the beginning.
The guidance specifies that the clinical protocol should address what happens when manufacturing fails or is delayed: whether a second leukapheresis and manufacturing attempt will be made, whether the patient will be replaced by additional enrollment, and what bridging therapy — treatment administered to manage disease progression during the manufacturing window — will be offered and how it will be handled analytically.
The bridging therapy question is more complex than it initially appears. Patients who receive bridging therapy between leukapheresis and CAR-T infusion may have different baseline disease status at the time of infusion than patients who do not. If the clinical trial does not pre-specify separate analysis of bridged and unbridged patients, interpreting the efficacy signal becomes substantially more difficult. The guidance recommends reassessing baseline disease before lymphodepletion — the conditioning regimen administered before CAR-T infusion — and conducting pre-specified separate analyses for all patients, bridged patients, and unbridged patients. This is not a retrospective analysis option. It is a prospective study design requirement.
The manufacturing failure contingency plan also has enrollment and timeline implications. If replacement enrollment is part of the contingency plan, the trial must be designed and powered to accommodate a defined failure rate. The failure rate must be estimated from early-phase experience — which means early-phase trials have the additional function of generating data to calibrate the manufacturing failure assumption used in later-phase power calculations. Developers who treat early-phase trials as purely safety and dose-escalation exercises miss this function.
The Commercial Manufacturing Implication
The manufacturing control challenges described in the CAR-T guidance are not temporary problems that resolve at commercial scale. They are structural features of autologous cell therapy that persist from IND through BLA and into commercial manufacturing. The six approved autologous CAR-T products — Kymriah, Yescarta, Tecartus, Breyanzi, Abecma, and Carvykti — all manufacture patient-specific lots from patient-specific leukapheresis material. All of them face the starting material variability, lot-to-lot consistency, and logistical complexity that the guidance addresses.
What changes at commercial scale is the operational burden of managing these challenges across a larger patient volume, a distributed manufacturing network, and a supply chain that must coordinate leukapheresis scheduling, material shipment, manufacturing slot allocation, product release, and clinical site administration for each individual patient simultaneously across potentially hundreds of patients at dozens of sites.
Programs that build robust manufacturing control infrastructure during clinical development — defined CPPs, in-process controls, validated comparability frameworks, chain of identity systems — enter commercial manufacturing with a significantly stronger operational foundation than those that treat CMC as a regulatory compliance function rather than a clinical development enabler.
Manufacturing control is not the variable that gets discussed when CAR-T clinical results are presented. It is the variable that determines whether those results are achievable at all.
MKA Insights works with CAR-T developers to build manufacturing strategy, CMC frameworks, and regulatory execution plans that reflect the operational realities of autologous cell therapy. If your program is navigating the transition from early clinical development to late-stage or commercial manufacturing, we bring the cross-functional perspective that these decisions require.
MKA Insights