CAR-T cell therapy

The starting point of CAR-T manufacturing. The patient's own blood runs through a device that skims off only the "mononuclear cells" containing the T cells used as raw material.

Blood flows through a continuous-flow centrifuge, which returns red cells and plasma to the body and continuously collects only the target cell layer. This gathers the required amount while keeping the burden on the patient low.

The number and vigor of the T cells collected here largely decide how good the final product will be. That is why this foundational step is the one where standardization matters most.

Mononuclear cells, including the T cells that are the raw material for CAR-T, are collected from the patient's own blood. The quality of the starting material (T cell number, composition and fitness) largely determines whether the final product succeeds, so standardizing this step is the top priority.

Alternatives in practice

• Intermittent-flow centrifugal apheresis · Terumo BCT COBE Spectra (older-generation mononuclear cell collection by two-stage chamber separation) — For sites still running legacy equipment. Replacement with the Spectra Optia successor is under way, and new installations are not recommended.
• Intermittent-flow apheresis · Fresenius Kabi Amicus Separator (MNC collection function) / Haemonetics MCS+ — Widely used in transfusion departments. Choose to match the site's existing platform.
• Downstream separation from a whole blood draw (research / small scale) · Whole blood bag collection + the density gradient separation device from [1] (Sepax C-Pro, etc.) — For research use with patients who cannot undergo apheresis or with small samples. Not suitable for clinical production scale.

The "prep work" that removes the bystanders in the collected blood that get in the way of proliferation. This pretreatment step concentrates only the mononuclear cells.

Centrifugation that exploits density differences, or washing, separates platelets, red cells and plasma. For frozen material, the DMSO in the cryoprotectant is also washed out after thawing.

These contaminants cause T cell aggregation and suppress proliferation. Removing them first keeps the selection and activation that follow consistent.

The apheresis product contains platelets, red cells and monocytes, which cause T cell aggregation, suppress proliferation, lower transduction efficiency and interfere with in-process testing. Isolating and debulking the mononuclear cells by closed-system density gradient centrifugation or by washing and concentration, and, for frozen material, washing out the cryoprotectant (DMSO) after thawing, is standard pretreatment before the subsequent selection and activation.

Alternatives in practice

• Counterflow centrifugal elutriation · Terumo BCT Elutra Cell Separation System (size separation by elutriation) — Removes platelets and monocytes by fractionation without Ficoll. Closed system, well suited to large volumes.
• Spinning-membrane / filtration-based automated wash and concentration · Fresenius Kabi LOVO (handled by Cytiva; spinning-membrane filtration with a 4 µm track-etched membrane) / Cytiva Sefia — A simple route that skips density gradient separation and only washes, concentrates and removes platelets.

The "time buffer" between collection and manufacturing. The collected material is frozen, stored and shipped, and waits until its turn in manufacturing comes.

At the collection site, the material is frozen slowly at a constant rate together with a DMSO-containing protective solution. It travels through the cold chain and stays frozen until a manufacturing slot opens.

International shipping delays and slot waits are absorbed here. Frozen starting material has been shown to have quality equivalent to fresh material.

For commercial CAR-T, in both centralized and decentralized manufacturing, controlled-rate freezing and a frozen hold of the apheresis starting material have become standard to absorb international shipping delays from the collection site to the manufacturing site and waits for manufacturing slots. Multicenter studies have confirmed that frozen starting material often shows viability, expansion, CAR-positive rate and cytotoxicity equivalent to fresh material.

Alternatives in practice

• Fresh shipping without freezing · Validated 2–8 °C shipping container (e.g., temperature-controlled containers such as Pelican BioThermal Credo / Cytiva) — Skips freezing for short distances and immediate manufacturing. Time constraints are tight.
• LN2-free conduction-cooled controlled-rate freezing · Cytiva VIA Freeze (conduction-cooled CRF, no liquid nitrogen needed) — For sites without LN2 supply infrastructure and for decentralized manufacturing.

The step that "picks out" only the T cells, the main players, from the mixture. Magnetic beads select the cells that carry the marker.

Positive selection with magnetic beads, using CD3 on the T cell surface (and CD4/CD8 as needed) as the marker. Monocytes, granulocytes, red cells and others are left behind.

Higher T cell purity stabilizes the efficiency of the later activation and gene transfer. This final selection evens out the quality of the starting cells.

The collected material contains monocytes, granulocytes, red cells and other cells. Removing them and enriching T cells (CD3, and CD4/CD8 as needed) stabilizes the efficiency of activation and gene transfer.

Alternatives in practice

• Counterflow centrifugal elutriation · Terumo BCT Elutra (label-free separation by density and size) — Uses no antibody beads, avoiding their cost and residual-bead concerns. Purity is somewhat lower.
• Skip selection and activate PBMC directly · (No equipment needed) Add the activation reagent from [4] directly — Shortens the process. Variability in the cell composition of the final product increases.
• Naive/memory T subset selection · Miltenyi CliniMACS CD62L GMP MicroBeads / CD45RA depletion — A next-gen design that enriches Tscm/Tcm to improve persistence.

The "start switch for proliferation" that wakes up dormant T cells. In the resting state, neither division nor gene transfer makes progress.

When a reagent delivers the 2 activation signals, CD3 and CD28, at the same time, the T cells wake up and start dividing.

Awakened cells accept the virus more readily, so the efficiency of the subsequent CAR gene transfer rises sharply. This move decides how well the later steps work.

Resting T cells do not divide and are hard to transduce. Activating them with CD3/CD28 signals starts proliferation and greatly raises the efficiency of retroviral/lentiviral gene transfer.

Alternatives in practice

• Activation with paramagnetic beads · Thermo Fisher CTS Dynabeads CD3/CD28 (4.5 µm magnetic beads, product number 40203D) — Well established, but the beads must be removed with a magnetic stand after culture [7].
• Soluble nano-reagents / reversible multimers · Juno/Bristol Myers Squibb-type Expamer (Streptamer-type reversible MHC/antibody multimers) / Cloudz (R&D/Bio-Techne dissolvable hydrogel) — Aims for zero residual beads. Platform dependent.
• Artificial antigen-presenting cells (aAPC) · K562-based aAPC feeder cells (4-1BBL/CD64-expressing lines, custom-built by each company) — Strong co-stimulation is obtained, but feeder-derived impurities must be controlled.

The heart of CAR-T. This step writes the CAR gene, the "eyes" that find cancer cells, into the blueprint (genome) of the T cell.

Cells are infected with lentivirus carrying the CAR gene, and the gene is integrated into the genome. The cell then becomes a soldier that can strike its target precisely.

The keys are the transduction efficiency and the number of copies that enter 1 cell. These tie directly to later dose design and safety control.

The CAR gene that recognizes a cancer antigen is integrated into the T cell genome. This is the core step of CAR-T, and the transduction efficiency and the copy number per 1 cell are directly tied to dose and safety.

Alternatives in practice

• Gamma-retroviral vector transduction · RetroNectin (Takara Bio, recombinant fibronectin CH-296)-coated vessel + GMP retroviral vector — Kymriah uses lentivirus; Yescarta/Tecartus use gamma-retrovirus. Depends on dividing cells.
• Transposon (non-viral) gene transfer · Sleeping Beauty (piggyBac, etc.) transposon systems + MaxCyte GT/ExPERT electroporation — Avoids the cost of viral vector manufacturing. Delivery efficiency and copy number control take some engineering.
• Site-specific knock-in by CRISPR · MaxCyte ExPERT GTx / Lonza 4D-Nucleofector + AAV6 donor (TRAC locus, etc.) — Uniform expression and simultaneous gene disruption. For allogeneic/next-gen designs.

The step that expands the engineered T cells to "enough for treatment". They are cultured in large numbers cleanly in a closed, automated bioreactor.

Cells grow while nutrients and oxygen are maintained by perfusion (continuously flowing in fresh medium) or agitation. The device manages the conditions automatically and keeps manual handling to a minimum.

Simply increasing the number is not the goal. The quality goal is to avoid exhausting the cells by over-expanding them and to keep the "young" properties that favor engraftment.

Cells are expanded to the number required for therapy. At the same time, the quality goal is to avoid exhausting the cells through excessive proliferation and to keep a young memory phenotype that favors engraftment and persistence.

Alternatives in practice

• Static/semi-automated culture in gas-permeable vessels · Wilson Wolf G-Rex (gas-permeable silicone membrane flasks, 100 mL–5 L) — Low cost and low shear stress. Closed-system automation has to be added separately.
• Fully automated cartridge-based manufacturing · Lonza Cocoon Platform (single-use cassette) / Cytiva Sefia + Xuri — For decentralized and point-of-care manufacturing.
• Short-term culture / no culture (next-gen) · Miltenyi Prodigy shortened protocol (T cell culture shortened to a few days) / in vivo CAR systems — Advanced approaches that aim at Tscm enrichment and a shorter manufacturing time.

The "rinse" before formulation. This step washes away the reagents and residues used in culture and brings the cells to just the right concentration.

Cells are washed and concentrated in an automated device, and if they were activated with beads, the magnetic beads are also captured and removed. Nothing unnecessary is left behind.

Cleaning out the medium, cytokines and residues protects recovery and final quality. It is a cleanup step on the way to the finished product.

When beads were used for activation, the magnetic beads are removed and the medium, cytokines and residues are washed away. Adjusting the cells to an appropriate concentration before formulation protects recovery and final quality.

Alternatives in practice

• Washing by counterflow centrifugal elutriation · Terumo BCT Elutra — Size-based separation also removes dead cells, debris and platelets at the same time.
• Bead removal with a magnetic stand · Thermo Fisher CTS DynaMag (clinical magnetic stand) — Brings residual beads within specification in processes that use Dynabeads.
• Washing and concentration by membrane separation · Fresenius Kabi LOVO (spinning membrane) — High recovery; processes large volumes in a short time.

The step that puts the finished cells into product form and freezes them into a "deliverable state". It is storage to buy time before the cells reach the administration site.

Cells are frozen at a slow, constant rate together with a DMSO-containing protective solution, then filled. This secures a margin for transport and administration.

Freezing stress damages cells. Protecting them with a protective solution and a controlled rate is meant to keep post-thaw viability and function.

Cells are frozen so that they can be shipped to the administration site and the time until administration is secured. The aim is to protect the cells from freezing stress and keep post-thaw viability and function.

Alternatives in practice

• General-purpose, in-house prepared freezing solution containing DMSO · In-house solution of Cryoserv/Edwards DMSO + dextran 40 + human albumin — Low cost, but composition variability and standardization are issues.
• Passive freezing · Thermo Fisher Mr. Frosty / CoolCell (isopropanol-based and alcohol-free passive cooling containers) — For small scale and research use. Reproducibility of -1 °C/min is poorer.
• Immediate administration of a fresh (non-frozen) product · (No freezing equipment needed) Immediate transport in a validated 2–8 °C container — Only where nearby manufacturing and short-term stability can be ensured.

A check on "did we make it exactly as intended?" This step examines whether the marker called CAR appears on the cell surface.

Flow cytometry shines light on each cell, 1 cell at a time, and measures what percentage of all cells are CAR-positive. It shows whether the product is as designed.

This is the product's "proof of identity" and also the basis for designing the dose. It is an important quality attribute that supports both identity and dose.

Confirms that the CAR molecule is displayed on the cell surface as intended and what percentage of cells are CAR-positive. This is the identity of the product and an important quality attribute that also underlies dose design.

Alternatives in practice

• Protein L / anti-Fab generic detection · Biotinylated Protein L (Thermo Fisher 29997, etc.) + streptavidin fluorophore (BD/Thermo) — Detects diverse scFvs regardless of the target antigen. Specificity is somewhat lower.
• Anti-idiotype antibody detection · Construct-specific anti-idiotype antibody (commercial/in-house, PE-labeled) — Highest specificity, but a dedicated reagent is required for each CAR.
• Tag/reporter detection · Anti-Myc / anti-Strep-tag II antibodies, or antibody detection of co-expressed EGFRt/tCD19 (e.g., cetuximab-based anti-EGFR) — A tag must be designed into the construct.

A safety test that measures whether "too much gene was put in". It accurately counts how many copies of the virus entered 1 cell.

A high-precision method called ddPCR quantifies the copy number of the integrated vector as an absolute value. The count can be made with little variability.

Too many copies raise the risk of tumorigenesis through insertional mutagenesis. As a general guide, the upper limit is controlled at an average of 5 copies or less.

Measures how many copies of the virus are integrated in 1 cell. Too many copies raise the risk of tumorigenesis through insertional mutagenesis, so an upper limit is controlled as a regulatory safety indicator (generally an average of 5 copies/genome or less is the guide).

Alternatives in practice

• Real-time quantitative PCR · Thermo Fisher QuantStudio 5 / Bio-Rad CFX Opus + standard curve — Simpler than ddPCR, but inferior in absolute quantification accuracy and low-copy reproducibility.
• CAR-positive rate by flow cytometry as a surrogate indicator · Clinical flow cytometer from [9] — A rapid surrogate for transduction efficiency. It cannot measure the copy number itself.
• Next-generation sequencing (integration site analysis) · Illumina MiSeq/NextSeq + LAM-PCR/S-EPTS — Evaluates insertion sites and clonality. Excessive for routine VCN QC.

A test that determines "how many living cells will be administered". It measures viable cell number and viability, and is central to the release decision.

Imaging or fluorescence automatically judges live or dead for each cell, 1 cell at a time. In particular it measures the number of living CAR-positive cells and the overall viability.

Viable cell number is the dose itself, and viability is an indicator of the soundness of the product. Both are pillars of the release specification and important values that decide pass or fail.

The number of viable cells administered (especially CAR-positive viable cells) is the dose itself. Viability indicates the soundness of the product and is central to the release specification.

Alternatives in practice

• Trypan blue dye exclusion (automated) · Beckman Coulter Vi-CELL BLU / Thermo Fisher Countess 3 FL — Simple and low cost. Discrimination of subtle dead cells is inferior to AO/DAPI.
• Simultaneous live/dead and CAR quantification by flow cytometry · Clinical flow cytometer from [9] + viability dye — Viability and expression are obtained simultaneously from 1 sample. Low throughput.
• Manual counting with a hemocytometer · Improved Neubauer hemocytometer + trypan blue — For backup/intermediate control. Large operator-to-operator variation.

A test of the "inner quality" of the cell population. It analyzes in detail what types of T cells are present and in what proportions.

Flow cytometry with simultaneous staining in multiple colors reads out the CD4/CD8 ratio, the memory-to-effector ratio and the degree of exhaustion (PD-1, etc.).

These are quality attributes related to engraftment, persistence and efficacy. They are used to reduce lot-to-lot variability and keep the product consistent.

Examines the CD4/CD8 ratio, the ratio of memory (naive/central memory) to effector cells, and exhaustion markers (PD-1, LAG-3, TIM-3). These are quality attributes related to engraftment, persistence and efficacy, and are used to control product consistency.

Alternatives in practice

• Spectral flow cytometry · Cytek Aurora / Sony ID7000 — High-dimensional analysis with more than 40 colors. Compensation is not needed, but unmixing is required.
• Mass cytometry (CyTOF) · Standard BioTools CyTOF XT (formerly Fluidigm Helios) + metal-labeled antibodies — 40+ parameters without fluorescence overlap. Throughput is limited and cost is high.
• Single-cell RNA-seq phenotyping · 10x Genomics Chromium + Illumina sequencer — Comprehensive transcriptome/CITE-seq. For characterization rather than QC.

A potency test to confirm "does it really work?" This step actually tests whether the product can truly attack cancer cells.

CAR-T cells are stimulated with the target antigen, and the assay measures whether they release cytokines such as IFN-γ and can kill cancer cells. Function is what is verified.

This is proof that the product has the "power to work". It is regarded as essential from a regulatory standpoint, and it is an indispensable test for assuring efficacy.

Confirms that CAR-T cells recognize the target antigen, actually kill cancer cells and release cytokines. This is proof of the product's "power to work" and an essential potency test from a regulatory standpoint.

Alternatives in practice

• Real-time cytotoxicity (impedance) · Agilent xCELLigence RTCA — Label-free, continuous acquisition of killing kinetics. Target cell adhesion is required.
• Flow cytometry/dye-based cytotoxicity assay · [9] flow cytometer + CFSE/CellTrace + viability dye — Quantifies the specific killing rate, which depends on the E:T ratio.
• Multiplex cytokine profiling · Bio-Techne/ProteinSimple Ella (automated multiplex ELISA) / Luminex MAGPIX — Measures IFN-γ/TNF-α/IL-2, etc. simultaneously to evaluate function from multiple angles.

A safety test that rules out "the virus having run out of control". It examines whether the virus used in manufacturing has regained infectivity.

The cells are first amplified in highly susceptible cells, and qPCR looks for the trace (provirus sequence) that would appear if a replication-competent virus were present.

If one were present, it would be a serious risk of insertional mutagenesis and viremia. This is a test required by regulation to show that it is "not there".

Confirms that the viral vector used in manufacturing has not regained the ability to self-replicate and become an infectious virus. Its presence would be a serious risk of insertional mutagenesis and viremia, so this is a safety test required by regulation.

Alternatives in practice

• PERT (product-enhanced reverse transcriptase) assay · qPCR-based PERT (e.g., SG-PERT, Bio-Rad/Thermo cyclers) — Detects RT activity with high sensitivity. Useful for general screening.
• p24 antigen ELISA (HIV-1 vector) · Takara/Clontech Lenti-X p24 Rapid Titer Kit / PerkinElmer HIV-1 p24 ELISA — An indicator of free particles/contamination in lentiviral systems.
• RCR testing for gamma-retroviral products · Extended S+L- focus assay / PG4 cells — Replication-competent virus testing for retroviral vector products.

A sterility test that confirms "no stray microbes got in". It matters because CAR-T is a living-cell product that cannot be sterilized afterward.

Using pharmacopoeial methods, the absence of bacterial and fungal contamination is confirmed by filtration through a filter or by direct inoculation. The product is tested as living cells.

Contamination would cause serious infection on administration. Because the shelf life is short, a rapid method whose results arrive in time is also required.

CAR-T is a living-cell product that cannot be terminally sterilized. Unless the absence of bacterial and fungal contamination is confirmed, administration can cause serious infection. A rapid method that fits within the short shelf life is also required.

Alternatives in practice

• Rapid sterility test using an automated blood culture system · bioMérieux BacT/ALERT 3D / BD BACTEC FX — A rapid method essential for short-shelf-life products. Equivalence validation against USP <71> is required.
• PCR/nucleic acid-based rapid microbial detection · Rapid molecular methods (e.g., each vendor's rapid micro method) — Results within hours. Equivalence verification against culture methods is a prerequisite.
• Gram staining (in-process control) · Light microscope + Gram stain kit — For in-process monitoring, not for release decisions.

A test aimed at contamination that is hard to see. It rules out mycoplasma, which is hard to notice in culture, and endotoxin, which causes fever.

Mycoplasma is detected by rapid PCR, and endotoxin is quantified by the kinetic turbidimetric/chromogenic LAL method. Both are tested quickly.

Both harm the patient when administered. Because the shelf life is short, rapid and kinetic methods are used to rule them out reliably before release.

Mycoplasma is a contaminant that is hard to see in culture and easy to introduce, and endotoxin is a pyrogen. Both are harmful when administered, so they are ruled out before release. Because the shelf life is short, rapid PCR and kinetic methods are used.

Alternatives in practice

• Pharmacopoeial mycoplasma culture method + indicator cell method · Hoechst staining (DNA-binding dye) + Vero/3T3 indicator cells, broth/agar culture — Classical method requiring 28 days. The comparison reference for establishing equivalence of rapid PCR.
• Gel-clot LAL · Lonza/Associates of Cape Cod gel-clot LAL reagent — Simple and low cost. Semi-quantitative and less rapid.
• Recombinant factor C (rFC) method · Lonza PyroGene rFC Assay / bioMérieux ENDONEXT EndoZyme II — Does not use horseshoe crab-derived LAL (addresses animal welfare). Kinetic fluorescence measurement.

A test of "are we bringing in anything extra?" It confirms that impurities mixed in during manufacturing are within the acceptance limits.

Remaining magnetic beads are counted, and residual DMSO, residual vector and culture-derived cytokines are quantified. The test checks that none exceeds the specified values.

Keeping these process-related impurities to a minimum leads to patient safety. It is the finishing safety check.

Confirms that residual activation beads, residual DMSO, residual vector, culture-derived cytokines and the like are at or below the specified values. This minimizes the impurities carried into the patient and assures safety.

Alternatives in practice

• Measurement of residual DMSO/additives · Agilent headspace GC-FID (7890/8890) / GC-MS — Solvent residues in the frozen product are controlled in the release specification.
• Quantification of residual cytokines/medium components · R&D Systems/Bio-Techne ELISA / ProteinSimple Ella — Quantifies carry-over of IL-2 and the like. For immunogenicity and specification compliance.
• Evaluation of residual vector (free particles) · Takara Lenti-X p24 / VSV-G qPCR (QuantStudio/QX200) — Risk assessment of contamination by free vector particles.