Last November we published POSTmark CAR-T: Delivering Therapies Without Delay, arguing that manufacturing continuity—not biology—is the rate-limiting step for most CAR-T programs. This year at ASGCT, just months later, it was abundantly clear that the conversation has shifted under everyone’s feet. In vivo CAR-T, once a distant technology, is now the most promising frontier in the quest to scale up cell and gene therapy, and the questions it raises are, once again, manufacturing questions.
Here is how fast the ground has moved, and what it means for the infrastructure and expertise that the field will need next.
The premise of in vivo CAR-T has always been simple: instead of extracting a patient’s T cells, engineering them in a facility, and reinfusing them weeks later, deliver the CAR construct directly into the body and let the immune system build the therapy itself. No apheresis. No lymphodepletion. No multi-week ex vivo manufacturing slot. The appeal is access, cost, and speed—reaching patients who can’t wait, or who were never candidates for autologous CAR-T at all.
What changed is that the capital has caught up with the concept. Over roughly the past year, big pharma has spent heavily to buy its way in. AstraZeneca acquired EsoBiotec for up to $1 billion in early 2025; Gilead’s Kite picked up Interius BioTherapeutics for $350 million and signed a Pregene Biopharma collaboration worth up to $1.6 billion; AbbVie bought Capstan Therapeutics for $2.1 billion; and Bristol Myers Squibb acquired Orbital Therapeutics for $1.5 billion. Eli Lilly entered twice in a single year—Orna Therapeutics for up to $2.4 billion, then Kelonia Therapeutics for $7 billion.¹² The pace has been brisk enough that analysts now talk less about whether pharma wants in and more about whether any independent targets remain to buy.³
Figure 1. In vivo CAR-T deal timeline. Pharma has spent billions in a single year, split across lentiviral and LNP/mRNA delivery; bubble size shows deal value. (Sources: Labiotech.eu, 2026; FierceBiotech, 2026)
The clinical story is moving in parallel, if more unevenly. First-in-human readouts across viral and non-viral platforms have been, in the words of one industry commentator, "a mixed bag."⁴ An early lentiviral in vivo study in advanced multiple myeloma showed encouraging responses, and a first global LNP-based study in systemic lupus erythematosus delivered proof-of-concept B-cell depletion.⁵,⁶ The signals are real, but cohorts are tiny and long-term follow-up is thin—this modality is at the very beginning of its evidence curve.
The shift isn’t arbitrary. It reflects a genuine inflection in delivery technology. As CPTx founder and CEO Hendrik Dietz framed it, the idea of programming T cells directly inside the patient has become realistic only recently, on the back of converging advances in viral-vector engineering and lipid-nanoparticle design.⁷ The economic logic follows from the biology: an in vivo therapy behaves more like a conventional biologic, where a single manufactured batch can, in principle, treat many patients—replacing a process in which every step is patient-specific and there is no economy of scale.⁷
It also expands what is clinically thinkable. Because duration of CAR activity becomes a tunable parameter—transient bursts for an autoimmune “reset,” durable expression for cancer—researchers can pursue repeat-dosing regimens and larger, non-life-threatening indications that were simply out of range when CAR-T meant a one-shot, permanent intervention.⁷ A recent regulatory perspective makes the same point from the manufacturing side: in vivo approaches could compress the production timeline to an estimated 1.5–3 weeks and lower projected cost per course by at least 50%, while shifting from a “scale-out” model to true “scale-up.”⁸
Figure 2. From bespoke to off-the-shelf: ex vivo vs. in vivo workflow comparison. In vivo CAR-T eliminates complex, expensive, and time-intensive steps like apheresis, patient lymphodepletion, and cell expansion. This allows a 2-4 week, patient-specific scale-out to shift to 1.5-3 week scale-up, extending a single batch to many patients. (Source: Lu et al., 2026)
Here is what often gets lost in the excitement around the biology: every in vivo delivery approach rests on a distinct, demanding manufacturing capability. The therapy is no longer a cell product engineered under controlled ex vivo conditions and characterized before release—it is the vector itself, administered directly to the patient. That single fact resets the quality, analytical, and safety bar.
Lentiviral vectors remain the most clinically validated route for stable CAR integration. For in vivo use they are engineered with targeting ligands—DARPins, nanobodies, or modified envelopes—to home in on T cells in circulation. Recent work has shown that retargeted, later-generation lentiviral vectors can generate high levels of CAR-T cells and durable B-cell depletion in humanized models.⁹ But producing targeted LVV at clinical scale, with the titer, purity, and potency required for direct administration, is not a simple extension of ex vivo vector manufacturing. The specifications are tighter, the analytical demands greater, and the tolerance for impurities narrower, because the vector is the drug rather than a reagent.⁸
LNP/mRNA platforms offer a different trade space. mRNA delivery is transient by design—attractive for safety and repeat dosing, but dependent on precise particle engineering for biodistribution and T-cell targeting. The depth of the CMC challenge here was on vivid display at ASGCT 2026, where Acuitas Therapeutics reported that aldehyde impurities in ionizable lipid raw materials are a primary driver of RNA-lipid adduct formation—undesired bonds that suppress mRNA translation, frequently escape standard purity assays, and directly degrade potency.¹⁰ In other words, the quality of an incoming lipid lot can quietly determine whether the finished therapy works. That is a manufacturing and raw-material-control problem before it is ever a clinical one.
Targeting and the liver problem. Both modalities share a central engineering goal: get the construct into T cells and keep it out of the liver. The same Acuitas data showed an extended-circulation LNP with DARPin targeting achieving a 10-fold reduction in liver expression and driving CAR expression on more than 60% of circulating CD8+ T cells in non-human primates, with deep B-cell depletion at low doses.¹⁰ Sana Biotechnology, presenting at the same meeting, reported that its CD8-targeted fusosome platform achieved cell-specific delivery and complete B-cell depletion in NHPs without lymphodepletion.¹¹ Different mechanisms, same hard requirement—and each demands its own analytical methods to prove targeting specificity.
Figure 3. Where the challenges live: a modality trade-off matrix. In vivo CAR-T isn't one manufacturing problem but several — each delivery route trades off durability, redosing, targeting, and CMC burden differently. (Relative assessment; informed by Lu et al., 2026 and Acuitas, 2026)
What this means in practice: in vivo CAR-T is not one manufacturing problem but several, defined by which delivery modality a program pursues and how far along it is. And because the field hasn’t consolidated around a single approach—the deal flow above spans lentiviral, LNP/mRNA, fusosome, and circular-RNA platforms—programs increasingly need partners fluent across more than one.
It would be a mistake to read this as a clean generational handoff. Novartis, whose T-Charge platform shortened ex vivo culture time to preserve younger, stem-like T cells, argues there is room for both approaches—and continues to evaluate in vivo as a next “stepwise” innovation rather than a replacement.¹² Ex vivo therapies remain the clinical bedrock; they have saved tens of thousands of patients and built the regulatory and operational scaffolding that in vivo development now leans on.⁷ Meanwhile, autologous innovation continues at the edges, including preconditioning-free dosing strategies that chip away at the same access barriers in vivo aims to remove.¹³
The honest read is that the field is broadening, not flipping. For a manufacturing partner, that is precisely the point: the programs most likely to succeed will move fluidly between modalities, and the infrastructure has to keep up.
Regulators Are Watching as Challenges Arise
The same properties that make in vivo CAR-T attractive raise the safety and regulatory stakes. Because CAR-T generation happens inside the body, precise control over transduction, expansion, and persistence is harder, shifting the risk profile toward off-target effects, immunogenicity, and—for viral vectors—theoretical vector-related responses.⁸ Early clinical data already hint at post-infusion immune reactions, including acute inflammation and isolated neurotoxicity.⁶,⁸
The bar rises further for non-oncology indications. When the population is autoimmune rather than end-stage cancer, the acceptable safety margin is far lower, and decades of expected survival amplify long-term risks. Regulators are responding by treating these as gene therapy products under accelerated-but-rigorous frameworks—RMAT in the U.S., PRIME in the EU—while pushing for early engagement, harmonized comparability standards, and long-term follow-up.⁸ Capital is flowing into exactly this space: CREATE Medicines recently closed a $122M Series B to advance a repeat-dose-capable mRNA-LNP CD19 program in autoimmune disease and oncology.¹⁴ The investment thesis and the regulatory caution are growing up together.
For developers, the practical implication is that the analytical and CMC package is no longer downstream paperwork—it is the product's regulatory backbone. Methods to quantify targeting efficiency, characterize conjugation and particle attributes, and detect impurities like the lipid adducts above must be built into the program from the first GMP run.
This is the backdrop against which an enabling partner earns its place. The capabilities that matter for the in vivo transition aren’t a single cell-therapy skill—they span payload, vector, and delivery, and they have to share a quality system.
Landmark Bio, an Artis BioSolutions company, was built around that kind of multi-modality coordination. Its GMP lentiviral vector platform runs at 50–200L scale on HEK293 suspension systems with third-generation plasmid systems and integrated two-stage chromatography purification, with analytical development and QC performed in-house by the same team supporting GMP—preserving method continuity from development through clinical supply. LVV and CAR-T manufacturing operate under one roof in modular cGMP suites supporting autologous and allogeneic workflows. Upstream, Artis also manufactures the synthetic DNA that underpins vector production, with GMP-ready starting material available in as little as two weeks, and brings AAV manufacturing alongside LVV for programs whose delivery strategy may evolve.
The argument from the November POSTmark piece applies with more force here: splitting vector and cell-therapy manufacturing across sites and quality systems creates scheduling, comparability, and timeline risk that compounds as a program scales. In vivo CAR-T, which demands tight coordination across payload design, vector engineering, delivery-system manufacturing, and analytical characterization, is the case where that integration stops being a convenience and becomes a strategic asset.
In vivo CAR-T has moved from preclinical promise toward clinical reality faster than most expected—propelled by billions in investment and genuine advances in delivery, tempered by small cohorts, real safety questions, and an evidence base still being written. The programs advancing through Phase 1 now will define the manufacturing and regulatory infrastructure the field needs next.
Whether a team is advancing an ex vivo program, evaluating a shift toward in vivo delivery, or building across multiple modalities, the manufacturing decisions made today will shape timelines, regulatory packages, and ultimately patient access for years.
Contact us to talk through how integrated capabilities across payload, vector, and delivery can support your program.
References
1. Labiotech. In vivo CAR-T cell therapy gains momentum as big pharma bets billions. labiotech.eu (2026).
2. FierceBiotech. After Lilly’s $7B Kelonia deal, are there any in vivo CAR-T biotechs left to buy? fiercebiotech.com (2026).
3. FierceBiotech. Has Novartis’ T-Charge been overtaken by in vivo CAR-Ts? Execs argue there’s room for both. fiercebiotech.com (2026).
4. Lowe, D. The Latest CAR-T Work. In the Pipeline, Science. science.org (2026).
5. Xu, J. et al. In-vivo B-cell maturation antigen CAR T-cell therapy for relapsed or refractory multiple myeloma. Lancet 406, 228–231 (2025).
6. Wang, Q. et al. In vivo CD19 CAR T-cell therapy for refractory systemic lupus erythematosus. N. Engl. J. Med. 393, 1542–1544 (2025).
7. Labiotech (interview with H. Dietz, CPTx). In vivo CAR-T cell therapy momentum. labiotech.eu (2026).
8. Lu, J. et al. The in vivo revolution in CAR-T therapy medicinal products: challenges and regulatory prospects. Signal Transduct. Target. Ther. 11, 192 (2026). doi.org/10.1038/s41392-026-02633-4
9. Coradin, T. et al. Efficient in vivo generation of CAR T cells using a retargeted fourth-generation lentiviral vector. Mol. Ther. 33, 4953–4967 (2025).
10. Acuitas Therapeutics highlights in vivo CAR T cell engineering and ionizable lipid quality attributes at the 2026 ASGCT Annual Meeting. BioSpace/Business Wire. biospace.com (2026).
11. Sana Biotechnology presents preclinical data for in vivo CAR T cell therapy SG293 in NHPs. biospace.com (2026). [Presented as ASGCT 2026 Abstract No. 20.]
12. FierceBiotech. Novartis CEO ‘continuing to evaluate’ in vivo CAR-Ts, but no deals in the works. fiercebiotech.com (2026).
13. The Medicine Maker. This Week’s CGT News: Cabaletta reports preconditioning-free CAR-T data in pemphigus vulgaris. themedicinemaker.com (2026).
14. The Medicine Maker. CREATE Medicines raises $122 million to advance in vivo CAR pipeline. themedicinemaker.com (2026).
About the Contributors