Cell and Gene Therapy Manufacturing: The Real Work Is Still the Workflow

The past week did not bring a clean breakthrough story in cell and gene therapy manufacturing. What moved was the layer under the headlines: change control, comparability, capacity planning, and the harder problem of proving a process still makes the same product after anything shifts.

What changed this week

The most concrete development in the record is still the FDA’s manufacturing change framework for CGT, which remains the pressure point for operators trying to modify process, facility, or scale without breaking comparability. FDA’s position is plain: manufacturing changes have to be handled through risk based quality systems, and changes that look like improvements can still produce a product that is not comparable.

That matters because the operational burden is not the science alone. FDA’s guidance emphasizes that sponsors need enough CMC information to assess impacts on identity, quality, purity, and strength, and that vector or process changes can delay studies or create shortages if comparability material is not preserved. In other words, the hard part is often not whether the biology works. It is whether the chain of evidence survives the change.

Where the pain sits

CGT manufacturing is a stack of brittle handoffs. Source material, identity checks, transport, process steps, in process analytics, release testing, and batch disposition all carry their own failure modes, and the data for each step often lives in different systems or in spreadsheets held together by humans.

FDA’s guidance makes the scale out problem explicit: manufacturers are expected to align development pace with manufacturing maturity, use risk management, and show that any expanded capacity still supports product identity, strength, quality, purity, and potency. That is not a paperwork exercise. It is a proof problem.

This is where senior engineering and R&D teams usually feel the friction most sharply. The public story keeps circling scientific novelty while operators are stuck on release complexity, sample traceability, and whether a late change in processing or facility setup breaks comparability. When this goes wrong, the failure is concrete: lost samples, delayed release, manual reconciliation, and a therapy schedule that slips because one handoff was not clean.

Why adoption is hard

Adoption of manufacturing software, orchestration tools, and chain of identity systems is slow because CGT does not behave like a single standardized production line. Each step has unique constraints, and each product can introduce a different set of acceptance criteria, starting material limitations, and release dependencies.

FDA’s comparability framework also raises the bar for digital tooling: if a change touches process, facility, or scale, teams need evidence that the new state still matches the old one on critical attributes and that any deviation can be defended scientifically. That means software cannot just automate tasks. It has to preserve traceability, lock down version control, and support a defensible audit trail.

Teams stall because the implementation burden is not symmetrical. The software may look straightforward in a demo, but the real system includes exception handling, site specific workflows, QC review, and all the little human workarounds that grow around missing integrations. If the platform cannot absorb those realities, people fall back to email, shared drives, and spreadsheet glue, which is exactly how chain of custody breaks in the first place.

What failure looks like

Failure in CGT manufacturing is not abstract. It looks like a sample that cannot be traced cleanly from collection to release, a release decision delayed because the data package is incomplete, or a comparability package that collapses because the process change was not documented tightly enough.

It also looks like the ugly middle layer operators know well: manual spreadsheet glue, rekeyed metadata, delayed handoffs between collection and manufacturing, and exceptions that have to be resolved by email instead of by system logic. Once a chain of identity or custody breaks, the cost is not just operational drag. It can mean a batch cannot be released, a study cannot move forward, or a scaled process cannot be defended.

What to watch next

The signal to watch is whether more vendors and CDMOs start attaching software changes to actual process controls rather than generic automation language. The useful announcements will be the ones tied to chain of identity, batch tracking, electronic release, comparability support, or capacity expansion with evidence that the workflow is less brittle than before.

For now, the message from the manufacturing side is consistent: CGT is still a systems problem. The science may be remarkable, but the bottleneck is whether the operation can keep its own records straight under pressure.

If you are seeing a different bottleneck in the field, compare notes. The interesting part is usually not the slogan, it is the handoff that failed.