How to Choose a CDMO for Cell and Gene Therapy

The short answer: Choosing a cell and gene therapy (CGT) CDMO is first a question of sub-modality — viral-vector production (AAV, lentiviral), autologous cell therapy (patient-specific, scaled out) and allogeneic cell therapy (off-the-shelf, scaled up) are three different manufacturing problems, and few CDMOs are genuinely strong in more than one. After sub-modality, weigh the two things that make CGT uniquely hard: potency and characterization analytics (the assays that prove a living or genetic product works are immature and program-specific) and logistics (cryopreservation, chain-of-identity and, for autologous therapies, a vein-to-vein timeline). Capacity for viral vector in particular is chronically constrained, so secure it early. Screen the site's GMP and inspection record as for any sterile product, and expect the regulatory framework (RMAT in the US, ATMP in the EU) to shape the whole program.

CGT is the least commoditised corner of manufacturing. The product is often a living cell or a viral particle carrying a gene; the process frequently is the product; and the analytics needed to prove it works are still maturing. That makes CDMO selection less about capacity and more about whether the partner has genuinely solved your sub-modality before — and it makes a mid-program change of partner exceptionally risky.

Sub-modality first: three different problems

The most common and expensive error is treating CGT as one capability. It is at least three:

• Viral vector manufacturing (AAV, lentiviral) — cell-culture-based vector production and a demanding purification and characterization train. This capacity underpins most gene therapies and many cell therapies, and it is the industry's tightest bottleneck.
• Autologous cell therapy (e.g. many CAR-T products) — a patient-specific process run once per patient, scaled out (many parallel small batches) rather than up, with chain-of-identity and a vein-to-vein clock as core constraints.
• Allogeneic cell therapy — an off-the-shelf product from a donor or line, scaled up like a more conventional bioprocess, with different economics and logistics.

A vector specialist is not automatically a cell-therapy manufacturer, and an autologous operation is not built like an allogeneic one. Match the CDMO to your exact sub-modality and ask for programs in it.

1. Verify the specific manufacturing capability

Within your sub-modality, probe the real capability:

• For viral vector: the production system, titers and yields on comparable constructs, and the purification and characterization to control empty/full capsid ratios and impurities.
• For autologous: the ability to run many parallel batches reproducibly with airtight chain-of-identity and chain-of-custody.
• For allogeneic: the cell-expansion process and the scale-up path to commercial volumes.

Evidence — programs run, clinical batches made — matters far more here than installed floorspace.

2. Weight potency and characterization analytics heavily

This is where CGT programs most often struggle. For a living or genetic product, the assays that define identity, purity and — hardest of all — potency are immature, slow to develop, and frequently product-specific. A CDMO that cannot support robust potency and characterization assay development will leave you unable to prove your product does what it claims, exactly when a regulator or a scale-up decision demands it. Treat analytical and assay-development depth as a primary selection criterion, not a support function — the same "measure it or you cannot file it" principle as in formulation and biologics, amplified.

3. Confirm logistics: cryo, chain-of-identity, and timelines

CGT logistics are part of the manufacturing problem:

• Cryopreservation and cold chain for cells and vectors, with validated handling.
• Chain-of-identity and chain-of-custody — for autologous products, an error here is a patient-safety event, not a paperwork slip.
• Vein-to-vein timeline for autologous therapies — the CDMO's turnaround directly affects patients waiting for treatment.

A partner strong in the bioprocess but weak in these logistics is not a safe autologous manufacturer.

4. Secure capacity early, and align on the regulatory path

Two practical realities:

• Capacity is scarce, viral vector especially. Credible CGT capacity is booked out, so the timeline conversation has to happen early — this is not a market where you can assume a second source exists, making the single-versus-dual-source question acute.
• The regulatory framework shapes everything — expedited pathways (RMAT in the US, the ATMP framework and PRIME in the EU) and their inspection expectations. Confirm the CDMO has taken a comparable product through the relevant path and check the site's FDA compliance record.

Common pitfalls

• Treating CGT as one capability and buying a vector specialist for a cell-therapy program, or vice versa.
• Under-scoping potency analytics, then being unable to prove the product at scale-up or review.
• Ignoring chain-of-identity and vein-to-vein for autologous products, where logistics is patient safety.
• Assuming capacity is available in a market where it is chronically constrained.

Where this fits

CGT is CDMO selection at its most specialised — sub-modality fit, potency analytics and logistics dominate, and capacity and regulatory strategy have to be planned unusually early. If you are still deciding what kind of partner you need, start with CRO vs CDMO vs CMO.

PharmaTek links CDMO capabilities to each company's GMP and FDA compliance record across a directory of 22,000+ pharmaceutical companies and manufacturers, so a cell- or gene-therapy shortlist can be built for sub-modality fit and screened for compliance at once.

Frequently asked questions

What is the most important factor when choosing a cell and gene therapy CDMO? Sub-modality fit. Viral-vector production, autologous cell therapy (scaled out) and allogeneic cell therapy (scaled up) are three different manufacturing problems, and few CDMOs are strong in more than one. Match the partner to your exact sub-modality and its track record, then weight potency/characterization analytics and logistics — the two things that make CGT uniquely hard.

What is the difference between autologous and allogeneic cell therapy manufacturing? Autologous therapy is patient-specific: a batch is made from and returned to one patient, so it is scaled out (many parallel small runs) with chain-of-identity and a vein-to-vein timeline as core constraints. Allogeneic therapy is off-the-shelf from a donor or line, scaled up like a more conventional bioprocess, with different economics and logistics. The two require differently built operations.

Why are potency assays such a challenge in cell and gene therapy? Because the product is a living cell or a genetic construct whose mechanism is complex, so the assays that prove identity, purity and especially potency are immature, slow to develop, and often product-specific. A CDMO without deep assay-development capability leaves you unable to demonstrate the product works when a regulator or a scale-up decision requires it — making analytics a primary selection criterion.

Is viral vector manufacturing capacity hard to secure? Yes. Viral vector (AAV, lentiviral) is the tightest capacity bottleneck in the sector and underpins most gene therapies and many cell therapies. Credible capacity is frequently booked well ahead, so timeline and slot conversations have to happen early, and you cannot assume a ready second source exists.

What regulatory pathways apply to cell and gene therapy manufacturing? Expedited and specialised frameworks shape CGT programs — RMAT (Regenerative Medicine Advanced Therapy) designation in the US and the ATMP framework plus PRIME in the EU — each with its own inspection and CMC expectations. Confirm the CDMO has taken a comparable product through the relevant pathway and check the manufacturing site's FDA and EU/UK GMP record.

---

Sourcing a cell or gene therapy CDMO? Explore the supplier directory or book a demo to screen candidates against live GMP and FDA compliance data.