Engineering Guide · Vendor-Neutral

In Vivo vs Ex Vivo CAR-T: Manufacturing, Delivery and Clinical Guide

In vivo (single IV dose reprogramming T cells in the body) versus ex vivo (apheresis, ex vivo engineering, reinfusion) CAR-T comparison IN VIVO Engineered in the body tLNP CAR-mRNA Single IV dose tLNP-mRNA or engineered lentivirus ~24-72 h to CAR expression Target COGS $2-10k / dose VS EX VIVO Engineered in a facility Apheresis Activate → Transduce → Expand → Fill CAR-T product Cell product delivered Autologous or allogeneic 3-6 weeks vein-to-vein COGS $100-150k / dose
Figure 1: In vivo CAR-T dosed as a single IV of targeted LNP or engineered lentivirus, reprogramming circulating T cells in the body. Ex vivo CAR-T removes T cells, engineers them in a facility, and reinfuses the finished cell product.
Quick Verdict

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

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 allogeneic

Global 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 vivo

Real-world use cases

Where clinical programmes are converging on one architecture or the other.

Autoimmune · Phase 1
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.

Hematologic malignancy · Phase 1
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.

R/R DLBCL · Approved
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.

Allogeneic · Phase 2
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 Planner

Cost and lifecycle considerations

Total cost of ownership: three tiers

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 shippingNone$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

Ex vivo CAR-T sponsors (approved products)

Frequently asked questions

What is the difference between in vivo and ex vivo CAR-T?
Ex vivo CAR-T removes T cells from the body, engineers them in a manufacturing facility (autologous or allogeneic), and reinfuses the finished cell product. In vivo CAR-T delivers the CAR transgene directly into circulating T cells inside the patient using a targeted lipid nanoparticle (LNP) or an engineered lentiviral particle; no cells ever leave the body. Autologous and allogeneic are both ex vivo. In vivo replaces the manufacturing plant with a single IV or SC dose.
Is in vivo CAR-T approved by the FDA?
As of August 2026, no in vivo CAR-T product has received FDA or EMA marketing approval. All seven approved CAR-T products (Kymriah, Yescarta, Tecartus, Breyanzi, Abecma, Carvykti, Aucatzyl) are ex vivo autologous. Multiple in vivo candidates are in Phase 1: Capstan CPTX2309 (CD8-tLNP + CD19 CAR mRNA), Umoja UB-VV111 (VivoVec lentivirus), Interius INT2104, and academic CD19-CD8-tLNP programmes. Regulatory frameworks for in vivo gene delivery are still evolving.
How does in vivo CAR-T work?
In vivo CAR-T delivers a CAR-encoding nucleic acid to T cells inside the patient. Two dominant platforms exist. Targeted LNPs (tLNPs) conjugate an anti-CD3, anti-CD5, anti-CD7, or anti-CD8 antibody or VHH to the surface of a lipid nanoparticle carrying CAR mRNA; the antibody triggers receptor-mediated endocytosis, mRNA is translated for days to weeks, and the CAR appears transiently on T-cell surfaces. Engineered lentiviral particles (Umoja VivoVec, Interius) pseudotype the envelope with a T-cell-directed binder (anti-CD3, anti-CD7) so the vector integrates only into T cells, giving durable CAR expression. Both platforms replace apheresis, ex vivo activation, transduction, expansion, and cryopreservation with a single IV dose.
How much does in vivo CAR-T cost compared to ex vivo?
Ex vivo autologous CAR-T carries a manufacturing cost of goods sold (COGS) of approximately $100,000 to $150,000 per patient dose and list prices of $475,000 to $525,000; total episode cost including hospital care can approach $2 million. In vivo CAR-T targets small-molecule-like COGS in the low thousands of dollars per dose because there is no per-patient cell engineering, no apheresis, no cryopreservation, and no cell-therapy suite. Published projections put in vivo COGS in the $2,000 to $10,000 range at commercial scale, though no approved product exists yet to confirm real-world numbers.
Does in vivo CAR-T avoid cytokine release syndrome (CRS)?
Not entirely, but early clinical data suggest lower CRS severity. LNP-mRNA in vivo CAR-T produces transient CAR expression (days to weeks) which caps the peak effector cell burden and shortens the toxicity window compared with the multi-year persistence of ex vivo autologous. First-in-human data from the CD8-tLNP CPTX2309 trial (autoimmune indication) and academic CD19-CD8-tLNP studies report no grade 3 or higher CRS. Lentiviral in vivo approaches (Umoja, Interius) generate durably integrated CAR-T cells and are more likely to reproduce the ex vivo CRS profile. CRS risk is delivery-platform-dependent, not intrinsic to in vivo.
How does in vivo CAR-T persistence compare to ex vivo?
tLNP-mRNA in vivo CAR-T produces transient CAR expression: the CAR appears within 24 to 72 hours of dosing, peaks over the first week, and decays over roughly 2 to 4 weeks as mRNA is degraded and CAR-bearing cells turn over. Lentiviral in vivo CAR-T (VivoVec, INT2104) integrates into T-cell genomes and can persist for months to years, mirroring ex vivo autologous. Ex vivo autologous CAR-T persists for months to years; ex vivo allogeneic typically persists for 1 to 3 months due to host-versus-graft rejection. Short persistence enables repeat dosing (a strength for autoimmune indications) and reduces long-term safety risk; long persistence is preferred where durable single-shot cure of oncology relapse is the goal.
What are the leading in vivo CAR-T companies?
As of August 2026 the in vivo CAR-T landscape is split into two camps by delivery platform. LNP-mRNA camp: Capstan Therapeutics (acquired by Lilly, CPTX2309 CD8-tLNP in Phase 1 for autoimmune indications), Orna Therapeutics (Merck partnership, circular RNA plus LNP), and academic groups running CD19-CD8-tLNP studies. Lentiviral camp: Umoja Biopharma (BMS partnership, UB-VV111 VivoVec platform in Phase 1 hematologic malignancies) and Interius BioTherapeutics (INT2104 lentiviral CAR in Phase 1). AbbVie and Eli Lilly are the dominant pharma sponsors after 2024 to 2025 M&A activity.
Which patients benefit most from in vivo versus ex vivo CAR-T?
Choose ex vivo autologous for aggressive relapsed/refractory hematologic malignancies where multi-year remission is the goal and the patient can wait 3 to 6 weeks for vein-to-vein manufacturing (DLBCL, ALL, multiple myeloma). Choose ex vivo allogeneic when time-to-treatment is decisive and off-the-shelf inventory beats a personalised product (consolidation, bridging, aggressive lymphomas). Choose in vivo (LNP-mRNA) when the indication tolerates or benefits from transient CAR expression: autoimmune diseases (lupus, myasthenia gravis, systemic sclerosis, RA), or where repeat dosing is preferable to a one-shot infusion. Choose in vivo (lentiviral) for oncology indications where durable persistence is required but manufacturing cost or logistics rule out autologous.

Resources and references