Cell & gene therapy manufacturing

• Autologous cell therapy — cells collected from the patient (apheresis), engineered (e.g. CAR transduction with a lentiviral or retroviral vector), expanded, formulated, cryopreserved and returned to the same patient. The lot size is one. Kymriah, Yescarta, Tecartus, Breyanzi, Abecma, Carvykti.
• Allogeneic cell therapy — from a healthy donor, scaled to many patient doses (off-the-shelf NK, iPSC-derived, gamma-delta T). Larger batches but tighter HLA / immunogenicity controls.
• In-vivo gene therapy — viral or non-viral vector delivered directly to the patient (AAV — Luxturna, Zolgensma, Hemgenix, Elevidys; lentiviral; mRNA-based).
• Ex-vivo gene therapy — cells modified outside the body then returned (Zynteglo, Skysona, Casgevy / Lyfgenia for sickle cell).
• Tissue-engineered products and gene-editing CRISPR / base / prime editors — most regulated as biologics with the same underlying CGMP expectations.

US — FDA CBER (Center for Biologics Evaluation and Research). The product is a biologic licensed under PHSA Section 351 (BLA) and regulated under 21 CFR 600 / 610 / 211. 21 CFR 1271 governs HCT/P donor eligibility + tissue establishment registration; CGT-specific FDA guidance covers CAR T cell development, sameness considerations, potency assays, accelerated approval pathways and chemistry-manufacturing-controls expectations.

EU — Regulation (EC) 1394/2007 brings ATMPs (gene-therapy, somatic-cell, tissue-engineered, combined) into a centralised EMA pathway via the Committee for Advanced Therapies (CAT). EU GMP Part IV is the ATMP-specific GMP guideline (replacing former Annex 2 ATMP sections); Annex 1 governs sterile manufacture; the EMA Pharmacovigilance system applies.

Accreditation — FACT (Foundation for the Accreditation of Cellular Therapy) and JACIE (Europe) standards apply to cellular-therapy clinical programmes and frequently flow into manufacturing-site expectations through commercial protocols.

For autologous CAR-T the patient's apheresis collection IS the starting material. A mix-up at any step delivers cells to the wrong patient — clinically catastrophic. Every CGT manufacturer therefore runs a fortified version of standard CoC:

• Patient ID + DIN (Donation Identification Number, ISBT 128 standard) on every container, label and electronic record from apheresis bag → cryobag → shipping shipper → infusion bag.
• Two-person verification at every transfer event (collection → ship-in → thaw → engineering → fill → ship-out → infusion). Electronic barcode + visual + e-signature.
• Closed processing systems (Miltenyi Prodigy, Lonza Cocoon, Cytiva Sefia) reduce open-handling opportunities for mix-up.
• ISBT 128 + DICOM cross-reference to clinical patient record.
• Final product label must be applied AND verified against the apheresis label and the patient identifier on the infusion order — the technologist scans both before release.

Vector production is its own CGMP discipline:

• Lentiviral / retroviral vectors — produced from a master cell bank (MCB) by transient transfection in HEK293 or stable producer cell lines, downstream-purified, characterised for VCN, titre, sterility, mycoplasma, RCL (replication-competent lentivirus) and endotoxin.
• AAV vectors — capsid serotype-specific (AAV2, AAV5, AAV8, AAV9, AAVrh10). Triple-transfection or producer-cell-line process, downstream affinity / iodixanol / CIM monolith purification, characterised for empty/full capsid ratio, titre, residual host-cell DNA + protein, in-vivo potency.
• ICH Q5A(R2) viral-safety, Q5D cell-substrate, Q5E comparability all apply.

Two recurring CGT scientific challenges:

• Potency — autologous CAR-T cannot use a static reference standard, so potency is established by functional assays (IFN-γ release, cytotoxicity, CAR positivity by flow, VCN by qPCR). FDA expects orthogonal methods that together capture mechanism of action.
• Comparability — process changes (closed-system upgrade, vector lot change, scale-up) require formal Q5E comparability protocols. CGT comparability is harder than mAb because the analytical methods themselves are less mature.

EU Annex 1 (2022) sterile-manufacture revision applies in full to ATMPs. US expectations are aligned through 21 CFR 211 + FDA aseptic-processing guidance. Typical CGT facility design:

• Grade A in BSCs / isolators / closed systems for open aseptic steps; Grade B background for open-handling rooms.
• Closed-system processing wherever possible — reduces Grade B footprint to ante-rooms only.
• Segregated suites per patient lot — autologous CAR-T sites often use single-use closed processing per patient with documented turnaround, OR fully isolated multi-patient bays.
• Air-handling — unidirectional flow at Grade A; pressure cascades; documented recovery times.
• Environmental monitoring — viable + non-viable particulate, surface contact plates, settle plates, personnel monitoring — at every critical operation.
• Cryopreservation — controlled-rate freezer; LN₂ vapour-phase storage; cryogenic shipping shipper (Cryoport-style) with dataloggers from −150 °C onwards.

CGT release tests typically cover:

• Identity — flow cytometry (CAR positivity, immunophenotype), molecular methods (qPCR for transgene, VCN).
• Purity — residual host cells, residual beads / cytokines, free / unencapsulated vector, residual transduction reagents.
• Strength / potency — viable cell count, transduction efficiency, functional assay.
• Safety — sterility (USP <71>), mycoplasma (USP <63> + rapid PCR), endotoxin (USP <85>), RCL (where viral-vector engineered).
• Many sterility methods take 14 days while the patient needs the dose now — most autologous CGT releases run rapid sterility (BacT/Alert, Bactec, Sigma Bioburdens) and ship under a conditional-release pattern with the 14-day USP <71> result as a post-shipment confirmation.

• Chain-of-identity verified only on paper — barcode mismatch goes undetected until cross-check at infusion.
• Vector lot release tests skipped or compressed under timeline pressure.
• Patient-specific batch release authorised without explicit two-person sign-off — Part 11 11.200 violation.
• Cryogenic shipper datalogger silent for the in-transit period; cold-chain excursion missed.
• Comparability protocol absent for a closed-system upgrade — entire historical dataset becomes unusable.
• Potency assay reduced to a single flow marker — FDA expects orthogonal methods.
• Environmental monitoring frequency relaxed without risk-based justification.
• FACT/JACIE accreditation lapsed during commercial scale-up.