Six Companies Are Racing to Replace the $400,000 Factory That Makes CAR-T Cells. Three Have Started Injecting Humans.
Traditional CAR-T therapy requires extracting a patient's T cells, engineering them in a cleanroom for four to six weeks, and infusing them back at one of roughly 200 specialized US centers. In vivo CAR-T skips every one of those steps: a single injection reprograms immune cells inside the body, and over $3.6 billion in pharma acquisitions says the bet is serious.
A single dose of Kelonia Therapeutics' KLN-1010 made CAR-T cells appear in multiple myeloma patients' blood without ever leaving their bodies. No leukapheresis. No cleanroom. No four-week wait. KLN-1010, a lentiviral vector injected directly into the bloodstream, received FDA clearance to expand its Australian trial into the United States in early 2026, making it the first in vivo CAR-T program to reach US clinical stage. Kelonia is not alone.
At the American Society of Gene & Cell Therapy annual meeting in Boston last week, Therorna Inc. presented data on TI-0032, a circular-RNA-based in vivo CAR-T aimed at autoimmune diseases and cancer, with a first-in-human trial already underway and US and China dual IND filings on track for this year. Meanwhile, AbbVie's $2.1 billion acquisition of Capstan Therapeutics last June secured CPTX2309, a lipid nanoparticle therapy that delivers CAR-encoding mRNA directly to circulating CD8+ T cells, which means the world's fourth-largest pharmaceutical company is now betting its autoimmune franchise on a technology that did not exist in clinical form three years ago.
Three human trials across three competing technology platforms, each built on a different molecular chassis, each pursuing the same radical premise: the factory is the problem.
The $700,000 Bottleneck
Traditional CAR-T therapy follows a baroque manufacturing chain. A hematologist extracts the patient's white blood cells through leukapheresis, and those cells ship overnight to a centralized GMP facility, where technicians use viral vectors to insert a chimeric antigen receptor gene. Engineered cells expand in bioreactors for two to four weeks, then ship back. The patient receives lymphodepleting chemotherapy to clear existing immune cells, then gets the infusion, then begins at least a week of inpatient observation for cytokine release syndrome and neurotoxicity, two potentially lethal complications that require ICU-level monitoring.
Cost? A systematic review of 53 studies found the mean drug price alone runs $391,060, representing 75% of total treatment costs. Add hospitalization, monitoring, and complication management, and real-world data puts the full cost at approximately $702,000 per patient for CAR-T versus $372,000 for bispecific antibody alternatives. Some centers report totals exceeding $1 million when ICU stays are factored in. A million dollars. For one patient.
Access is worse than the price tag suggests. Only about 200 US centers are qualified to administer CAR-T. A study published in Cancer Network found 60% of hematologic cancer patients live more than 30 minutes from the nearest qualified center. For rural patients, that gap stretches to hours. The Pharmacy Times documented median out-of-pocket costs of $5,547 and a mandatory care period exceeding four weeks, creating what researchers call "time toxicity" on top of the financial burden.
India offers a preview of what sheer manufacturing scale could accomplish even within the traditional ex vivo model: decentralized closed-system manufacturing has brought CAR-T costs to $25,000-$50,000, 85-93% cheaper than the American version of the same treatment.
Six Programs, Three Platforms
In vivo CAR-T companies have split into three technology camps. Each makes a fundamentally different bet about durability, safety, and manufacturing scalability, and the bets are mutually exclusive enough that the winner's technology platform will likely define the next generation of cellular therapy.
| Company | Platform | Target | Indication | Stage |
|---|---|---|---|---|
| Kelonia Therapeutics | Lentiviral vector | BCMA | Multiple myeloma | Phase 1 (US/Australia) |
| Therorna | Circular RNA + LNP | CD19 | Autoimmune/oncology | First-in-human |
| AbbVie (ex-Capstan) | Targeted LNP + mRNA | CD19 | Autoimmune | Phase 1 |
| Roche (ex-Poseida) | Allogeneic CAR-T | CD19+CD20 | MS, lupus | Preclinical/Phase 1 |
| Umoja Biopharma | Lentiviral (VivoVec) | Various | Oncology | Preclinical |
| Academic (Nature) | CRISPR + AAV | TRAC locus | B-ALL (mice) | Preclinical |
Viral vectors (lentiviral) permanently integrate the CAR gene into T cell DNA. Kelonia's approach mirrors traditional CAR-T's molecular mechanism but skips the external manufacturing step entirely, and Umoja Biopharma's VivoVec platform pursues a similar strategy with its own proprietary lentiviral design. The payoff, if it works, is enormous: one injection, permanent CAR expression, potentially durable for years without re-dosing. The risk is equally permanent: insertional mutagenesis, the same molecular accident that caused leukemia in early gene therapy trials for severe combined immunodeficiency in the early 2000s and nearly killed the field.
Non-viral lipid nanoparticles deliver mRNA (AbbVie/Capstan) or circular RNA (Therorna) that T cells translate into CARs. Expression is transient, lasting days to weeks depending on the RNA chemistry, but Therorna's circular RNA resists exonuclease degradation and buys a longer expression window than standard mRNA, which means fewer injections per treatment course and potentially lower cumulative immunogenicity. Redosable, no permanent genomic changes, lower theoretical cancer risk. But patients may need repeat injections, and the immune system can develop anti-LNP antibodies that chip away at efficacy with each subsequent dose, a problem the mRNA vaccine field documented extensively during COVID booster campaigns.
CRISPR-based integration offers permanent, site-specific insertion at the TRAC locus, the T cell receptor alpha constant gene. A study published in Nature demonstrated 19.7% TRAC-CAR editing in mice and complete responses in B-ALL tumor models using a dual-vector system with CD3-targeted delivery. Advantage: precision rivaling traditional CAR-T with no factory. Disadvantage: furthest from the clinic, with delivery efficiency in humans unproven.
The Cost Math
Here is where it gets interesting. Traditional CAR-T's $391,000 average drug price breaks down roughly as: viral vector manufacturing ($50K-$80K), GMP cell processing ($80K-$120K), quality control and release testing ($30K-$50K), logistics and cryopreservation ($15K-$25K), and the manufacturer's margin on a personalized biologic ($100K-$150K+). In vivo CAR-T eliminates the cell processing, logistics, and per-patient manufacturing entirely.
What remains? Drug substance manufacturing (LNPs or viral vectors produced in bulk), fill-finish, and distribution. These are batch-manufactured goods, not bespoke biologics, and the cost curve for batch biologics bends hard with volume: Moderna's COVID mRNA-LNP vaccines, the closest manufacturing analogue, cost $15-$37 per dose at hundreds-of-millions scale. CAR-T LNPs are more complex because they must target specific cell types, but the manufacturing paradigm is fundamentally the same: mix lipids, encapsulate RNA, fill vials, ship cold.
Conservative estimate: if drug substance costs $5,000-$15,000 per dose at modest scale (thousands of patients per year, not millions), and administration requires a single outpatient injection with 24-hour observation rather than weeks of inpatient care, total treatment cost could fall to $50,000-$120,000. That is an 83-93% reduction from current averages.
Wall Street agrees. AbbVie paid $2.1 billion for Capstan, and Roche paid $1.5 billion for Poseida's allogeneic CAR-T platform, a combined $3.6 billion in acquisition value for programs that have not yet produced pivotal trial data. Pharma companies do not write those checks for incremental improvements.
Limitations
Every number above carries substantial uncertainty, and that uncertainty compounds across every link in the chain. No in vivo CAR-T has completed a Phase 2 trial. Kelonia's Australian data is early, the patient count is small, and the efficacy signal is preliminary enough that the company has not disclosed response rates. Therorna's first-in-human trial has not reported efficacy data publicly. Cost projections for LNP manufacturing at therapeutic scale, as opposed to vaccine scale, remain theoretical because nobody has done it yet for this application.
Durability is an open question for LNP-based approaches. If patients need re-dosing every three to six months, the cumulative cost advantage narrows dramatically, and if anti-LNP immune responses after repeated dosing reduce efficacy with each injection, a known problem from mRNA vaccine boosters, the transient approach could prove more expensive than the permanent one it was designed to replace.
Lentiviral approaches carry a darker shadow: insertional mutagenesis led to leukemia in early gene therapy trials for severe combined immunodeficiency, an outcome that shut down the entire field for nearly a decade. Modern vectors have improved safety profiles, but the risk is nonzero and will require years of follow-up to quantify.
The Strongest Counterargument
Traditional CAR-T works, and it works spectacularly: six FDA-approved products have produced durable remissions in blood cancers that had no other treatment options, in patients who had exhausted every line of chemotherapy and were running out of time. That manufacturing infrastructure, however clunky, exists, produces therapies that save lives today, and has a regulatory framework that agencies already know how to evaluate.
In vivo CAR-T introduces a new and uncomfortable variable: you cannot control which cells get transduced, and unlike ex vivo manufacturing, where technicians verify that the right cells express the right receptor at the right level before reinfusing them, injecting a viral vector or LNP into the bloodstream and hoping it finds the right T cells is a controlled detonation. If off-target transduction of non-T cells triggers unexpected toxicity, the field could suffer a setback that delays all in vivo programs, not just the one that failed.
Kelonia's decision to skip lymphodepleting chemotherapy, a prerequisite for all six approved CAR-T products, is simultaneously the most exciting and most precarious design choice. Lymphodepletion clears competing immune cells to give engineered T cells room to expand. Without lymphodepletion, in vivo-generated CARs may be outnumbered and outcompeted by the patient's existing immune cells, which is the central unanswered question: does the approach produce enough functional CAR-T cells to matter?
What You Can Do
If you or a family member has been told CAR-T is the next step: Ask your oncologist about in vivo trials. Kelonia's KLN-1010 is recruiting in the US for relapsed/refractory multiple myeloma (ClinicalTrials.gov). AbbVie's CPTX2309 and Therorna's TI-0032 are enrolling for autoimmune indications. Clinical trial enrollment is free, often includes monitoring that exceeds standard of care, and may provide access to therapies years before commercial approval.
If you are an oncologist at a community practice: These programs could bring cellular therapy to your patients for the first time. Monitor ASGCT and ASH conference data for KLN-1010 efficacy readouts, expected in late 2026.
If you work in health policy or insurance: Start modeling now. If in vivo CAR-T reaches approval by 2029-2030, the eligible patient population expands from tens of thousands (current CAR-T) to potentially millions (autoimmune diseases alone). Current benefit designs are not built for this.
The Bottom Line
CAR-T therapy is the most powerful weapon in cancer medicine's arsenal, a treatment that can produce complete remissions in patients who have exhausted every other option, and most patients who need it will never receive it. The manufacturing chain is too slow, too expensive, too geographically concentrated, and too dependent on a bespoke production model that treats every patient like a custom order. Six companies are now attempting to solve all three problems simultaneously by eliminating the factory from the equation. Three have injected humans. None have proven it works at scale. But $3.6 billion in pharma acquisitions and three first-in-human trials within twelve months suggest the bet is serious. If any one of these approaches produces durable remissions from a single injection, the implications extend far beyond oncology: autoimmune diseases, organ transplant rejection, and conditions nobody has tried to treat with cellular therapy because the economics never made sense. The factory is the bottleneck. The race to remove it has started.