Download the Report
In-vivo cell therapy moves CAR-T manufacturing inside the patient
Buyers have committed about $21 billion in potential deal value to in-vivo cell therapy, which engineers T cells and other immune cells inside the patient, even though only one program has reached Phase 2. Sleuth’s third-quarter 2026 analysis counts 332 in-vivo programs, 246 of them CAR-T, and 20 deals since 2021. The promise is an off-the-shelf therapy dosed like a biologic, without apheresis, weeks of manufacturing or the lymphodepleting chemotherapy that ex-vivo CAR-T requires.
Two delivery architectures lead, and they tell different stories. Integrating lentiviral vectors aim for the lifelong durability of ex-vivo CAR-T and lean toward oncology. Transient RNA approaches delivered by lipid nanoparticle (LNP) express the CAR for days to weeks, can be redosed and fit an autoimmune reset. Together they account for about 63% of the pipeline and nearly all disclosed deal value, and Lilly bought one of each. The trade-off is control: ex-vivo cells can be tested and rejected before infusion, but an in-vivo CAR-T population is made inside the patient, cannot be inspected and depends on the patient’s own T cells. In-vivo therapy sits between CAR-T’s depth and T-cell engagers’ convenience, and it has to prove it can deliver both.
332 programs, one in Phase 2 and a field split between China and the US
CAR-T dominates, and China edges the US on volume
Of the 332 programs, 258 are preclinical, 2 are IND-enabling, 63 are in Phase 1, 8 in Phase 1/2 and one in Phase 2. CAR-T accounts for 246 (74%), followed by CAR-macrophage and other myeloid designs (22), multi-lineage and stem-cell approaches (19), CAR-NK (9) and TCR-T (6), with 30 programs in other categories. China originates 161 programs (48.5%) and the US 141 (42.5%), a more even split than in most modalities Sleuth covers.
The field is largely an off-the-shelf follow-on to ex-vivo CAR-T, and with a single Phase 2 program it remains an early bet on optionality.
Oncology leads, and autoimmune disease is the emerging channel
Of the 332 programs, 177 target oncology only, 53 autoimmune disease only and 49 both, with 53 in neither, so 102 (30.7%) touch autoimmune disease, including lupus, rheumatoid arthritis and systemic sclerosis. CD19 appears in 91 programs and is the only or primary target in 54, BCMA appears in 43 (15 as the primary target) and CD20 in 18 (9 as the primary target), mirroring the ex-vivo field. Sixty-two programs carry two or more targets.
CD19 and BCMA concentrate the near-term field, so multi-target and solid-tumor programs mark the open ground.
Delivery splits evenly between integrating viral and transient RNA
Lentiviral delivery accounts for 107 programs (32.2%), tilted toward oncology at 90 to 21. LNP-delivered mRNA or circular RNA accounts for 101 (30.4%), with 54 in oncology and 40 in autoimmune disease. Lentiviral integration copies the CAR into every daughter cell, giving durable expression that cannot be switched off and carries a theoretical insertional risk, and anti-vector immunity can block redosing. Non-integrating RNA is translated for days and then clears, so response depth can be titrated and redosed, and circular RNA lengthens the window. Eight further programs pursue permanent integration without a virus.
Dealmaking has run ahead of clinical validation
The 20 in-vivo deals since 2021 include 8 acquisitions or options and 12 licenses or collaborations, with about $12.2 billion committed to integrating-viral platforms and $9.0 billion to transient RNA. Disclosed value rose from $2.2 billion across four deals in 2024 to $6.6 billion across seven in 2025 and $12.0 billion across four in 2026 through the third quarter.
Lilly made the two largest, acquiring Kelonia (viral, up to $7.0 billion) in April 2026 and Orna (RNA, up to $2.4 billion) in February 2026. AbbVie acquired Capstan for up to about $2.1 billion in 2025, moving an in-vivo CD19 program into its immunology franchise, and Gilead, AstraZeneca, BMS and Johnson & Johnson have also bought in. Totals were disclosed for only 12 of the 20 deals and upfronts for 10, so the figures understate committed capital.
Owners of B-cell franchises are buying in-vivo platforms to secure their own biology at lower cost and to defend against off-the-shelf rivals.
The case for in-vivo is logistical first and clinical second
Off-the-shelf supply and lymphodepletion-free dosing are the unlock
Autologous ex-vivo CAR-T takes about 60 days from vein to vein, and even Carvykti’s manufacturing turnaround of about 30 days runs across more than 142 treatment centers. In-vivo therapy ships and doses like a biologic, and because it is designed to skip lymphodepletion, it could open outpatient and earlier-line use. Programming unmanipulated T cells in place may also avoid the exhaustion that ex-vivo culture imposes, and preclinical work shows in-vivo delivery can reach T cells in the spleen, a pool apheresis cannot draw on.
Both delivery arms have early human proof
Sleuth highlights seven programs that have reported human data. MagicRNA’s HN2301, an RNA approach, drove complete circulating B-cell depletion and cut lupus disease activity (SLEDAI) by up to 20 points within three months in five patients, without lymphodepletion, as reported in The New England Journal of Medicine. Kelonia’s viral KLN-1010 produced MRD-negative responses in all treated myeloma patients reported, LB2501 reached 100% response at its second dose level in lymphoma, and Starna’s STR-P004 was redosed eight times in lupus. ESO-T01, a viral myeloma program, produced MRD negativity in 4 of 5 patients but reported one death and grade 3 or higher adverse events in every patient.
With human data on both arms, the open question has shifted from whether in-vivo works to where its profile wins, and safety depends on the construct and dose rather than the category.
The patient becomes the bioreactor
In-vivo removes manufacturing failure but adds a new failure mode: the patient’s own T-cell fitness. Sleuth grades five requirements. Cell targeting is demonstrated in humans with CD8-targeted LNPs and T-cell-tropic lentivirus. Dose and expansion are emerging, with HN2301 producing 60% CD8-positive CAR-positive T cells and GT801 reaching up to 93% reprogramming. Safety is mixed. Persistence is open, because lentivirus integrates for durability while transient RNA does not yet, and off-target control is the core unsolved problem and the real differentiator between platforms.
T-cell fitness is weakest in the heavily pretreated cancers in-vivo most wants to treat, which favors earlier-line and autoimmune use.
In-vivo CD19 has to match CAR-T depth with engager convenience
The cleanest proof for in-vivo CD19 so far is in autoimmune disease, where HN2301 programmed the patient’s own polyclonal T cells rather than manufacturing them. The signal is early and indication-dependent. Efficacy hinges on transduction, biodistribution and T-cell fitness, which is worse after chemotherapy or anti-CD20 therapy. A program must clear B cells in marrow and lymph nodes, not just blood, and transient RNA may under-deliver the durability blood cancers need. In autoimmune disease, in-vivo CD19 competes with off-the-shelf T-cell engagers and next-generation B-cell depleters, a field Sleuth maps in the in-vivo CD19 competitive map and its T-cell engager report.
Beyond CAR-T, and China’s role in the field
One in six programs reaches beyond CAR-T, chasing solid tumors
Fifty-three of the 332 programs (16%) reach beyond CAR-T, led by CAR-macrophages, which can traffic into solid tumors CAR-T cannot reach, kill by phagocytosis and turn a cold tumor hot. Across all 332 programs, 44 (13.3%) name a solid tumor. The named clinical readouts Sleuth cites for these cell types come from ex-vivo or allogeneic programs, so in-vivo beyond CAR-T has little clinical proof of its own.
China is broad, early and holds the only Phase 2 trial
China’s 161 programs span about 66 companies. Sixty-three percent are preclinical and 34% touch autoimmune disease, and the field’s only Phase 2 trial is Chinese. China-originated programs also account for 23 of the 36 forward catalysts Sleuth tracks (64%). Across all 36, 25 are trial starts or first dosing and 11 are data readouts. China’s edge is translational speed and autoimmune mRNA-LNP delivery.
Umoja’s UB-VV400, Sana’s SG293, Capstan and AbbVie’s CPTX2309 and Starna’s STR-P004 are due in the second half of 2026, followed by Kelonia and Lilly’s KLN-1010 in the first half of 2027 and Vivacta’s GT801 in the second half. The next decisive readouts will test whether early proof turns into a differentiated profile against incumbent modalities.
What the in-vivo findings mean for developers and buyers
These implications are Sleuth’s interpretation of the data above.
For developers, choose the architecture for the indication. Durable, integrating constructs fit cancers that need deep, lasting responses, while transient, redosable RNA fits a drug-free autoimmune reset where safety margin matters most. Either way, the differentiator buyers will diligence is targeting and off-target control, and the patient populations with the fittest T cells are the best place to prove it.
For buyers, underwrite control, not just convenience. Deal values are running ahead of clinical validation, and the risks have moved from the manufacturing suite into the patient. Diligence should weigh how precisely a platform reaches T cells, how durable its expression is and how it behaves in patients whose immune systems have been depleted by prior therapy.
What Sleuth expects in in-vivo cell therapy
These predictions are Sleuth’s analytical interpretation, looking beyond what the data shows today.
- The field will split further by indication. Autoimmune disease will reward transient, redosable RNA for fast, drug-free B-cell resets, while oncology will still need durable, integrating constructs, and platforms carrying both toolkits will be best placed to straddle.
- Owners of B-cell and immunology franchises will keep buying in. Johnson & Johnson has already moved with Sail in autoimmune disease, and Novartis and Roche, each defending a major B-cell or immunology franchise, are the likeliest next buyers.
- More Chinese deals are coming, mostly as platform licenses built on translational speed and autoimmune mRNA-LNP delivery, with MagicRNA and Starna the companies to watch.
How Sleuth built this analysis
Pipeline scope
The analysis covers 332 in-vivo cell therapy programs as of the third quarter of 2026, meaning T cells or other immune cells genetically programmed inside the patient through viral, LNP or other delivery. Programs are classified by highest development phase, engineered cell type and originator geography, and clinical stage means Phase 1 or later.
Indications, targets and delivery
Indication classes are mutually exclusive: oncology only, autoimmune only, both, or neither. Target counts show every program that mentions a target and, separately, those where it is the only or primary target.
Deals
The deal set covers 20 deals from January 2021 through the third quarter of 2026 in which the acquired or licensed asset is an in-vivo cell therapy. Ex-vivo, Treg, delivery-only and gene-editing deals are excluded. Values are “up to” totals, upfronts were disclosed for 10 deals and totals for 12, and seven disclosed no economics, so figures are floors.
Clinical evidence and catalysts
Human data come from company disclosures, medical meetings and journals through the third quarter of 2026. Ex-vivo figures are precedent benchmarks, not in-vivo results, and the evidence grading is Sleuth’s interpretation for diligence. The catalyst calendar covers 36 forward events and is a floor rather than a census.
Interpretation
The implications and predictions are Sleuth’s analytical interpretation rather than direct findings from the data.
Pipeline, deal and clinical figures are as of the third quarter of 2026 unless otherwise noted.
Frequently asked questions about in-vivo cell therapy
What is in-vivo CAR-T?
In-vivo CAR-T delivers the genetic instructions for a chimeric antigen receptor directly into the patient, using a viral vector or lipid nanoparticle, so the patient’s own T cells become CAR-T cells in the body. It aims to avoid apheresis, weeks of manufacturing and lymphodepleting chemotherapy.
How many in-vivo cell therapy programs are in development?
Sleuth tracked 332 in-vivo cell therapy programs as of the third quarter of 2026, 246 of them CAR-T. China originates 161 and the US 141, and only one program has reached Phase 2.
What is the difference between lentiviral and LNP in-vivo CAR-T?
Lentiviral vectors integrate the CAR into the T cell’s genome for durable expression, which suits cancer but cannot be switched off. LNP-delivered RNA expresses the CAR for days to weeks and can be redosed, which suits autoimmune resets. Sleuth counted 107 lentiviral and 101 LNP-RNA programs.
How big are in-vivo cell therapy deals?
Twenty deals have committed about $21 billion in potential value, $12.2 billion to viral platforms and $9.0 billion to RNA. Lilly’s acquisitions of Kelonia (up to $7.0 billion) and Orna (up to $2.4 billion) in 2026 are the largest.
Has in-vivo CAR-T worked in patients?
Early data are promising. MagicRNA’s HN2301 depleted circulating B cells and reduced lupus disease activity by up to 20 points in five patients without lymphodepletion, and viral programs such as KLN-1010 have produced MRD-negative responses in myeloma. Safety varies by construct, and durability is unproven.
What role does China play in in-vivo cell therapy?
China originates 161 of 332 programs across about 66 companies, holds the field’s only Phase 2 trial and accounts for 64% of forward catalysts. Its strengths are translational speed and autoimmune mRNA-LNP delivery.
