In Vivo vs Ex Vivo CAR-T: Manufacturing, Delivery and Clinical Guide
Autologous and allogeneic are both ex vivo. The real fork is in vivo (no cells leave the body) versus ex vivo (cells engineered in a facility). In vivo CAR-T dosed as a targeted LNP with CAR mRNA is transient, cheap, repeat-dosable, and best fit for autoimmune indications. Ex vivo autologous remains the standard for durable single-shot oncology response.
Key differences at a glance
- In vivo CAR-T: Single IV dose of a targeted LNP or engineered lentivirus reprograms T cells inside the patient. No apheresis, no facility, no cryobag.
- Ex vivo CAR-T: T cells removed by apheresis, engineered in a manufacturing suite (autologous or allogeneic), released, cryopreserved, and reinfused.
- Cost difference: Ex vivo autologous COGS is $100-150k per dose; in vivo targets $2-10k per dose at commercial scale (no approved product yet).
- Best for autoimmune (lupus, MG, SSc): in vivo LNP-mRNA (transient CAR, repeat dosable, no lymphodepletion for tLNP autoimmune programmes).
- Best for R/R DLBCL, ALL, multiple myeloma: ex vivo autologous (seven approved products; multi-year persistence for durable remission).
Side-by-side comparison
| Factor | In vivo CAR-T | Ex vivo CAR-T (autologous) |
|---|---|---|
| Where engineering happens | Inside the patient's bloodstream | In a cell-therapy manufacturing suite |
| Delivery format | Targeted LNP + CAR mRNA, or T-cell-pseudotyped lentivirus, given IV or SC | Cryopreserved cell product, thawed and infused IV |
| Vein-to-vein time | Minutes to hours (single dose) | 3-6 weeks (autologous), 5-7 days (allogeneic off-the-shelf) |
| Apheresis required | No | Yes (autologous) or donor apheresis (allogeneic) |
| CAR persistence | Days to weeks (tLNP-mRNA); months to years (lentiviral) | Months to years |
| Repeat dosing feasible | Yes (tLNP-mRNA) | Difficult (one manufactured product per campaign) |
| Manufacturing COGS per dose | $2,000-10,000 (projected) | $100,000-150,000 |
| Regulatory status (Aug 2026) | Phase 1 only (Capstan, Umoja, Interius, academic) | 7 approved products (Kymriah, Yescarta, etc.) |
| Best-fit indication | Autoimmune, chronic dosing, high-volume oncology | R/R hematologic malignancies (DLBCL, ALL, MM) |
Values reflect typical published specifications and current-as-of-August-2026 clinical status. Your programme's data and regulator's current guidance take precedence.
In vivo CAR-T in detail
In vivo CAR-T therapies deliver the CAR transgene directly into circulating T cells inside the patient's body, replacing the entire ex vivo manufacturing chain with a single infusion. There are two dominant delivery platforms: targeted lipid nanoparticles (tLNPs) carrying CAR mRNA, and engineered lentiviral particles pseudotyped to bind T cells specifically. Both approaches are in Phase 1 clinical trials for hematologic malignancies and, increasingly, autoimmune indications.
How it works
The tLNP-mRNA route — championed by Capstan Therapeutics (acquired by Eli Lilly in 2025) and Orna Therapeutics (Merck partnership) — conjugates an anti-CD8, anti-CD5, or anti-CD7 antibody or single-domain VHH to the surface of a lipid nanoparticle encapsulating mRNA that encodes a CAR (typically anti-CD19 or anti-BCMA). The antibody handle steers the LNP to the target T-cell subset; receptor-mediated endocytosis delivers the mRNA cargo; ribosomes translate the CAR for 5 to 14 days; the reprogrammed T cells attack CAR-target-bearing malignant or autoimmune B cells. When mRNA is degraded and CAR-bearing cells turn over, the therapy resolves. Capstan's lead candidate CPTX2309 (CD8-tLNP encapsulating CD19 CAR mRNA) entered Phase 1 in June 2025 for B-cell-mediated autoimmune disorders. Academic groups have also published a first-in-human CD19-CD8-tLNP study in relapsed/refractory B-cell lymphoma with no grade 3+ CRS across four dosed patients.
The engineered lentiviral route — championed by Umoja Biopharma (BMS partnership; VivoVec platform) and Interius BioTherapeutics — pseudotypes the lentiviral envelope with a T-cell-directed binder (typically anti-CD3 or anti-CD7) so the vector transduces only T cells after IV infusion. Because lentivirus integrates into the genome, the resulting CAR-T cells persist and self-replicate like conventional ex vivo autologous CAR-T. Umoja's UB-VV111 and Interius's INT2104 are both in Phase 1 trials for CD19-driven hematologic malignancies. Combined analysis of early lentiviral in vivo CAR-T dosing reported four MRD-negative responses across ongoing trials in 2026.
When in vivo CAR-T wins
In vivo CAR-T wins on time, cost, access, and repeat dosing. A single IV can be prepared in a community oncology or rheumatology clinic; no cell-therapy accreditation, no apheresis suite, no cryoshipper. Manufacturing COGS is target-adjacent to a small-molecule biologic ($2-10k projected per dose versus $100-150k for autologous). For autoimmune indications where the therapeutic goal is to reset B-cell tolerance rather than achieve permanent CAR persistence, transient tLNP-mRNA expression is arguably a feature: the drug clears, safety exposure ends, and the physician can redose if relapse occurs. In vivo is also the only realistic path to CAR-T at the scale of the ~1 million lupus patients or ~10 million rheumatoid arthritis patients worldwide.
Ex vivo CAR-T in detail
Ex vivo CAR-T is the incumbent architecture. Every approved CAR-T product — Novartis Kymriah, Gilead/Kite Yescarta, Tecartus, BMS Breyanzi, BMS/2seventy Abecma, J&J/Legend Carvykti, and Autolus Aucatzyl — is manufactured ex vivo. The autologous variant uses the patient's own T cells; the (still-investigational) allogeneic variant uses gene-edited donor cells. Both remove cells from the body, engineer them in a facility, and reinfuse the finished product.
How it works
An autologous ex vivo campaign starts with leukapheresis to collect ~10 billion PBMCs from the patient. The apheresis product is shipped cold or cryopreserved to the manufacturer. In the facility, T cells are enriched (typically by CD4/CD8 bead selection), activated with anti-CD3/anti-CD28 beads or soluble activator, transduced with a CAR-encoding lentiviral or retroviral vector at MOI 1-5, expanded for 7-14 days in a G-Rex or WAVE bioreactor, formulated in a cryobag, released against product-specific specifications, and shipped back to the treatment centre. The patient receives lymphodepleting chemotherapy (typically fludarabine/cyclophosphamide) 3-5 days before infusion. Total vein-to-vein time is typically 3-6 weeks; novel decentralised platforms have compressed this to 10-14 days.
Allogeneic ex vivo uses healthy donor apheresis as starting material and adds gene edits — TRAC knockout to disable the endogenous TCR (preventing graft-versus-host disease), B2M knockout to reduce HLA class I surface expression, and sometimes CIITA and CD52 knockouts. Edits are made with TALEN (Cellectis), CRISPR-Cas9 (Caribou Biosciences, CRISPR Therapeutics), or base editing. The edited product is cryopreserved as inventory, thawed on demand, and shipped to treatment centres with 5-7 day vein-to-vein time. Allogene's cema-cel is the most advanced allogeneic candidate.
When ex vivo CAR-T wins
Ex vivo autologous wins on regulatory precedent and clinical durability. Seven approved products means clear FDA and EMA guidance on CMC, potency assays, comparability, and release specifications. Multi-year persistence and multi-year remissions in DLBCL, ALL, and multiple myeloma set a bar in vivo has not yet matched in oncology. Where the target patient population is small enough that per-dose economics are secondary to peak response rate and durability — R/R aggressive hematologic malignancies — the manufacturing burden is worth carrying. Ex vivo also wins where the target requires cell-intrinsic engineering that current in vivo delivery cannot replicate at scale: multiplex knockouts, TCR knock-in via HDR, armored CAR designs with logic gates.
Pros and cons
In vivo CAR-T
Advantages
- No apheresis, no manufacturing facility, no cryoshipper — a single IV replaces a 3-6 week supply chain.
- Projected COGS of $2-10k per dose (versus $100-150k autologous), enabling scale at autoimmune population sizes.
- Transient CAR expression (tLNP-mRNA) caps peak toxicity and enables repeat dosing on relapse.
- Can be prepared in a community clinic — no cell-therapy accreditation, no dedicated CAR-T infusion pathway required.
Disadvantages
- Zero approved products; regulatory framework for systemic in vivo gene delivery is still being written.
- Off-target transduction risk: lentivirus in bloodstream may reach non-T cells; tLNPs can accumulate in liver and spleen.
- Transient tLNP-mRNA CAR may not match the multi-year persistence needed for durable oncology remission.
- Immunogenicity of LNP components and pseudotyped envelopes can limit repeat dosing effectiveness.
Ex vivo CAR-T
Advantages
- Seven approved products; validated CMC, potency, and release-testing frameworks with FDA and EMA.
- Multi-year CAR persistence and multi-year remissions in R/R DLBCL, ALL, and multiple myeloma.
- Full control of cell dose, phenotype, and engineering — enables complex multi-edit and armored CAR designs.
- Off-target risk minimized: transduction happens in the manufacturing suite, not in the patient.
Disadvantages
- 3-6 week vein-to-vein time (autologous); many patients progress or die during manufacturing.
- COGS of $100-150k per dose and total episode cost approaching $2M; not viable at autoimmune population scale.
- Requires specialized cell-therapy infrastructure — apheresis suite, cryoshipper, accredited infusion centre.
- Manufacturing failure rate of 5-10% (out-of-spec product, insufficient expansion, contamination).
Which should you choose?
Match the delivery architecture to the indication, persistence requirement, and target population size.
Autoimmune indication (lupus, MG, SSc, RA)
Population is 10× to 100× larger than R/R oncology. Transient B-cell depletion followed by immune reset is the therapeutic goal; multi-year CAR persistence is not required and may increase harm.
Choose in vivo (tLNP-mRNA)R/R DLBCL, ALL, multiple myeloma
Durability of complete response is the outcome that matters. Regulatory pathway is settled. Autologous is the standard of care; allogeneic where time-to-treatment dominates.
Choose ex vivo (autologous)Time-to-treatment is the constraint
Aggressive lymphoma with weeks-median survival, or bridging not available. Off-the-shelf allogeneic (5-7 days) beats autologous (3-6 weeks); in vivo (hours) beats both.
Choose in vivo or allogeneicGlobal access at scale
LMIC deployment, community-clinic delivery, or payer economics require sub-$50k per-dose economics. Autologous cannot get there; in vivo is the only architecture that can.
Choose in vivoReal-world use cases
Where clinical programmes are converging on one architecture or the other.
Capstan CPTX2309 (CD8-tLNP + CD19 CAR mRNA)
In vivo, systemic IV, no lymphodepletion in autoimmune arm. B-cell-mediated autoimmune disease; Phase 1 initiated June 2025. Lilly acquisition 2025 marked the platform's mainstreaming.
Umoja UB-VV111 (VivoVec lentivirus, CD19 CAR)
In vivo, T-cell-pseudotyped lentivirus for stable CAR integration mirroring ex vivo autologous persistence. Dosed in hematologic malignancies; MRD-negative responses reported 2026.
Gilead/Kite Yescarta (axicabtagene ciloleucel)
Ex vivo autologous; anti-CD19 CAR with CD28 costimulation; list price ~$460k; standard of care 2L+ in aggressive lymphoma. Multi-year remission in ~40% of responders.
Allogene cema-cel (TRAC/CD52-edited donor CAR-T)
Ex vivo allogeneic; off-the-shelf inventory; ALPHA3 pivotal Phase 2 in first-line consolidation for LBCL. Primary EFS readout mid-2028. No allogeneic CAR-T is FDA-approved as of Aug 2026.
Sizing a cell therapy programme — autologous, allogeneic or in vivo?
The Cell Therapy Planner walks through apheresis yield, dose sizing, batch cadence, and manufacturing footprint so you can model an ex vivo campaign against an in vivo scenario before committing to a platform.
Open the Cell Therapy PlannerCost and lifecycle considerations
Manufacturing COGS is only the visible line. Ex vivo autologous also carries per-patient apheresis, cryoshipping, hospital cell-therapy accreditation, lymphodepletion, and inpatient CRS/ICANS monitoring. In vivo shifts almost all of that to a single infusion. Where an autologous CAR-T episode approaches $2M all-in, an in vivo dose is projected to look more like a standard biologic infusion at $10-30k total.
The dominant cost driver for ex vivo autologous is per-patient bespoke manufacturing: every dose requires its own apheresis, its own bioreactor cycle, its own release panel, its own cryoshipment. Fixed manufacturing overhead does not amortise across patients. Allogeneic amortises across doses but adds gene-edit complexity and inventory management, and no allogeneic product has yet cleared regulatory approval.
In vivo CAR-T pushes toward small-molecule-like batch economics: one GMP LNP or lentiviral batch can supply thousands of doses. The remaining costs are commercial (formulation, fill-finish, distribution) and clinical (IV infusion, short observation). If in vivo CAR-T achieves and sustains the current safety profile at scale, per-dose economics could drop by an order of magnitude relative to autologous.
| Cost component | In vivo (projected) | Ex vivo autologous (actual) |
|---|---|---|
| Manufacturing COGS per dose | $2,000-10,000 | $100,000-150,000 |
| Apheresis and shipping | None | $5,000-15,000 |
| Lymphodepletion (Flu/Cy) | None (autoimmune) or standard (oncology) | Required |
| Inpatient monitoring (CRS/ICANS) | Reduced (transient CAR) | 7-14 days standard |
| Total episode cost estimate | $10,000-30,000 (projected) | $500,000-2,000,000 |
In vivo cost figures are commercial-scale projections from academic and industry economic modelling; no in vivo CAR-T is yet approved. Ex vivo autologous figures reflect published payer and hospital data for approved products.
Vendor landscape
The commercial in vivo CAR-T field splits cleanly by delivery platform. Ex vivo autologous is dominated by the seven approved-product sponsors.
In vivo CAR-T sponsors
- Capstan Therapeutics (Eli Lilly): CD8-tLNP + CD19 CAR mRNA. Lead candidate CPTX2309 in Phase 1 for B-cell-mediated autoimmune disease. Acquired by Lilly in 2025 for ~$2B.
- Umoja Biopharma (BMS partnership): VivoVec engineered lentivirus platform. UB-VV111 in Phase 1 for hematologic malignancies. Also partnered with AbbVie on the VivoVec platform.
- Interius BioTherapeutics: T-cell-directed lentivirus. INT2104 in Phase 1 for CD19-driven hematologic malignancies. Independent player.
- Orna Therapeutics (Merck partnership): Circular RNA plus LNP delivery. Preclinical / early clinical stage for in vivo CAR programmes; up to $3.5B Merck deal signed 2023.
Ex vivo CAR-T sponsors (approved products)
- Novartis: Kymriah (tisagenlecleucel), first FDA-approved CAR-T (2017); B-ALL and DLBCL.
- Gilead/Kite: Yescarta (axi-cel), Tecartus (brexu-cel). Dominant DLBCL market share.
- BMS / 2seventy bio: Breyanzi (liso-cel), Abecma (ide-cel). Also partnered with Umoja on in vivo VivoVec.
- Johnson & Johnson / Legend Biotech: Carvykti (cilta-cel). BCMA CAR-T for multiple myeloma; the highest efficacy CAR-T MM product to date.
- Autolus Therapeutics: Aucatzyl (obe-cel). FDA-approved 2024 for adult R/R B-ALL.
- Allogene Therapeutics: Cema-cel (allogeneic CD19 CAR-T). Most advanced allogeneic candidate; ALPHA3 pivotal Phase 2.
Frequently asked questions
What is the difference between in vivo and ex vivo CAR-T?
Is in vivo CAR-T approved by the FDA?
How does in vivo CAR-T work?
How much does in vivo CAR-T cost compared to ex vivo?
Does in vivo CAR-T avoid cytokine release syndrome (CRS)?
How does in vivo CAR-T persistence compare to ex vivo?
What are the leading in vivo CAR-T companies?
Which patients benefit most from in vivo versus ex vivo CAR-T?
Resources and references
- The in vivo revolution in CAR-T therapy medicinal products: challenges and regulatory prospects — Signal Transduction and Targeted Therapy, May 2026. Regulatory framework review for in vivo CAR-T; covers CMC, potency assays, and comparability.
- From ex vivo to in vivo chimeric antigen T cells manufacturing: new horizons for CAR T-cell based therapy — Journal of Translational Medicine, 2024. Comprehensive review of delivery platforms and clinical translation challenges.
- First-in-human in vivo CAR T-cell generation using a CD8-targeted lipid nanoparticle platform in relapsed/refractory B-cell lymphoma — Journal of Clinical Oncology, 2026. First-in-human CD19-CD8-tLNP clinical data with safety and PD readouts.
- In vivo generation of CAR T cells: biology, delivery platforms, clinical promise, and translational challenges — Blood Immunology and Cellular Therapy (ASH), 2025. Peer-reviewed cross-platform comparison including LNP and lentiviral approaches.