One Patient. One Batch. No Second Chances: Why CGT Manufacturing Determines Regulatory Fate

In autologous cell and gene therapy, the manufacturing process isn't a step on the path to clinical outcomes—it is the clinical outcome. When a CAR-T batch fails, you can't re-collect a lymphoma patient's T-cells after they've progressed through another line of chemotherapy. When an adeno-associated virus (AAV) production run yields predominantly empty capsids, you can't blend it with a better lot or re-run the process while the patient waits. The therapy either reaches the patient, or it doesn't, and that determination is made on the manufacturing floor long before anyone opens an investigational new drug (IND) submission.

The Hold Problem Everyone Misunderstands

Cell and gene therapy (CGT) trials account for approximately 40% of all U.S. Food and Drug Administration (FDA) clinical holds despite representing only about 2% of trials.¹ That disproportion usually gets cited as evidence that CGT development is inherently risky and that holds are an unavoidable cost of working at the frontier of medicine.

That framing is both partially true and dangerously incomplete.

A peer-reviewed analysis of 33 publicly disclosed CGT clinical holds between 2020 and 2022 found that roughly 70% were triggered by adverse events—patient immune reactions, unexpected toxicities, fatal events during early-phase trials.² These are biological realities that reflect the genuine unknowns of first-in-class therapies that no manufacturing system can prevent. But the same analysis found that 21% of holds were triggered by CMC deficiencies—chemistry, manufacturing, and controls failures—and 9% by preclinical data gaps.³ Taken together, roughly one in three CGT clinical holds traces back to something that happened (or didn't happen) before the first patient was ever dosed.

Those are the holds we want to address. Not because safety holds don't matter but because the CMC holds are the ones that were preventable, and they carry consequences that compound in ways the industry consistently underestimates.

Among the 585 CGT INDs submitted between 2021 and 2023, approximately one in five were placed on clinical hold within the first 30 days of FDA review; programs stopped before they started because the manufacturing story couldn't withstand scrutiny.⁴ And once a CMC hold is issued, it takes an average of 8.4 months to resolve — 75% longer than protocol amendments and 29% longer than adverse event holds.⁵ For most CGT programs, that's not a regulatory inconvenience, it's an existential threat.

Why CMC Holds Happen

The industry tends to treat CMC holds as mysterious regulatory black boxes—opaque agency decisions that descend without warning. They aren't. They happen because CGT manufacturing has failure modes that traditional pharmaceutical manufacturing doesn't, and most programs enter development without building systems that account for them.

The Constraint You Can't Work Around

In autologous CGT, the fundamental constraint is irreversibility. You have one patient's cells, one manufacturing run, and no fallback. The traditional pharmaceutical model—make a batch, test it, investigate problems, make another batch—doesn't apply when the starting material is a specific patient's cells collected at a specific point in their disease course. By the time quality review is complete, the product has either been infused, shipped to a clinical site, or degraded past its stability window. Post-hoc investigation is an autopsy, not a correction.

For allogeneic products and viral vectors, the batches are larger and the biological starting material is more consistent, but the underlying principle holds: biological starting materials are irreplaceable, timelines are compressed, and every production run carries consequences that extend directly to patients.

That constraint doesn't just complicate manufacturing operations, it invalidates the quality review model that the broader pharmaceutical industry was built on.

The Variability You're Expected to Understand

Biological variability in CGT isn't a defect, it's inherent to the biology, and the FDA doesn't expect you to eliminate it. What the FDA expects is that you can distinguish between variability that is a natural consequence of working with living cells and biological materials, and variability that reflects a manufacturing process that isn't under control.

Consider what that variability looks like across modalities:

• In autologous CAR-T manufacturing, one patient's T-cells may expand 10-fold where another's expand 100-fold—same process, same protocol, completely different outcome.⁶ Your release specifications must hold across both and demonstrating that the difference is biology, rather than a process you don't control, is where the CMC argument either stands or falls.
• In AAV manufacturing, a production run that looks successful on paper may yield 70% to 90% empty capsids—and "good" and "poor" batches from the same process can differ by orders of magnitude.⁷ The FDA doesn't object to empty capsids existing; they object to sponsors who can't show that their process reliably delivers a product that actually works.
• In lentiviral manufacturing, the most common ways to measure your vector don't tell you how much of it works. Physical assays count particles; they can't tell you which ones are functional. The result is that your titer looks fine on paper while your actual potency is unknown. Swap measurement methods between Phase 1 and Phase 2 without data connecting the two, and you've created a comparability gap that the FDA will find before you do.

What Failures Look Like in a Submission

By itself, none of the above variability is disqualifying, the FDA's 2020 CMC guidance for gene therapy INDs explicitly describes a flexible, risk-based framework that allows for phased data submissions calibrated to clinical phase, permits proportionate comparability data for manufacturing changes, and acknowledges that early-stage programs will operate with less complete manufacturing characterization than late-stage programs.⁸ That flexibility was further reinforced in January 2026, when the FDA announced a broader clarification of its flexible CMC approach across clinical development, commercial specifications, and process validation.⁹ But it comes with a caveat: Phase 1 manufacturing systems need to have been capturing the right data from Day 1, because the baseline you're comparing against in Phase 2 and Phase 3 is built from the records you generated at the beginning.

When that baseline doesn't exist, the flexibility the FDA offers becomes inaccessible and the CMC holds that result tend to cluster around five failure patterns:

1. Potency assay gaps. You can measure vector genome copies or cell counts, but you can't demonstrate that those numbers correlate to the product working in a patient. You're manufacturing and releasing product without a meaningful signal of whether the batch will do what it's supposed to do.
2. Comparability failures. You moved from a 10-liter Phase 1 process to a 200-liter Phase 2 process. Without the Phase 1 baseline data—consistently captured and completely documented—you have no reference point from which to argue that the larger-scale product is the same thing that showed clinical benefit in Phase 1.
3. Traceability gaps. For autologous products, the inability to trace the finished product back to the specific patient's cells is a direct patient-safety failure. For allogeneic or vector products, it means that when something goes wrong downstream, you can't do root-cause analysis because the chain of custody was never systematically maintained.
4. Stability failures. Your product degrades during shipping or storage under real-world logistics conditions, and you don't have data demonstrating consistent potency at the point of patient administration. The FDA can't approve a product whose clinical performance at the patient's bedside can't be characterized.
5. Undocumented changes. Between Phase 1 and Phase 2, the manufacturing team optimized the process—adjusted media formulations, compressed culture timelines, switched to a higher-quality reagent supplier—but none of those changes were formally documented or run through change control. The FDA now sees a submission that describes a different process than the one that generated the Phase 1 clinical data, with no comparability package bridging the gap.

These are manufacturing execution failures that happen in the same places, for the same reasons, across programs—and they're the most expensive category of holds to resolve.

Why CMC Holds Are Irreversible in Ways Other Holds Aren't

An adverse event hold is painful, but the path forward is navigable. You investigate the event, update the protocol, adjust eligibility criteria or monitoring requirements, and submit a response. The timeline damage is real, but the foundation of your manufacturing program remains intact.

A CMC hold is structurally different. The path forward isn't a protocol amendment—it's reconstructing a foundation that was never built correctly in the first place. Generating comparability data you should have captured two years ago. Establishing the potency baseline you should have defined before your first clinical batch. Proving to the FDA that a process you've been running for months is under control, using data that doesn't exist because the systems that would have captured it were never implemented.

The data that wasn't collected can't be reconstructed. That's why CMC holds take an average of 8.4 months to resolve—not because the regulatory process is slow, but because sponsors are rebuilding their evidentiary foundation from scratch while the clock runs and patients wait.

For an autologous program, that wait isn't abstract: patients who were scheduled for treatment either hold on, if their disease allows, or exhaust other options and become ineligible. They don't return to the enrollment pipeline. Manufacturing slots go dark. Financing timelines built around clean clinical progression collapse. And the loss—of patients, capital, and competitive position—isn't recoverable.

What the Rest of This Blog Series Covers

Seventy percent of CGT clinical holds stem from adverse events—the biological unknowns that come with developing first-in-class therapies in seriously ill patients. Those are largely unavoidable. But the 21% driven by CMC deficiencies aren't. They happen because sponsors built systems designed for traditional pharmaceutical development and later discovered that those systems can't handle CGT.

The failure patterns are consistent across programs: potency assays that can't demonstrate the product works, comparability arguments that fall apart because Phase 1 data was never clean enough to serve as a baseline, process changes that outpaced the change control system and were never formally documented.

These are predictable failures that happen in the same places for the same reasons—which means they're preventable. The next post in this series identifies where CGT manufacturing breaks down, why those breakdowns happen, and what it takes to build systems that catch them before an FDA reviewer does.

References

1. "Reducing The Number Of Clinical Holds On Cell And Gene Therapies: Approaches For Sponsors And The FDA," Sandy Kweder and Sean Hilscher, Cell & Gene, March 25, 2024.
2. "Clinical holds for cell and gene therapy trials: Risks, impact, and lessons learned," Carolyn A. Wills, Daniela Drago, and Robert G. Pietrusko, Molecular Therapy – Methods & Clinical Development, Oct. 20, 2023.
3. Supra note 2.
4. "Cell & Gene Therapies: FDA Regulatory Considerations In 2024," World Pharma Today, accessed April 9, 2026.
5. Supra note 2.
6. "The Challenge of Variability in Chimeric Antigen Receptor T cell Manufacturing," Andrew D. Fesnak, Regenerative Engineering and Translational Medicine, Aug. 19, 2019.
7. "Methods to reduce empty capsid content in AAV manufacturing at scale," Patsnap Eureka, Sept. 2, 2025.
8. "Chemistry, Manufacturing, and Control (CMC) Information for Human Gene Therapy Investigational New Drug Applications (INDs)," FDA Guidance for Industry, Jan. 31, 2020.
9. "Flexible Requirements for Cell and Gene Therapies to Advance Innovation," FDA website, content current as of Jan. 11, 2026.