A Rocket level transaction made the long silent Aibo Biology stand in the spotlight again and let the new TCE species behind the transaction come to the front of the stage.
The core pipeline involved in this transaction, ABO2203, is the world's first CD19 × CD3 T cell adapter encoded and expressed in vivo using mRNA, which has entered clinical practice.

Simply put, it replaces the TCE bispecific antibody that was originally intended to be produced in vitro and then reinfused into the human body with one that can be manufactured by the human body based on the mRNA blueprint.

Novartis is willing to pay such a price for this early-stage clinical asset - a down payment of $575 million and a potential milestone payment of up to $7.2 billion - not only for the asset itself, but also for the "in vivo pharmaceutical" logic it represents.

In the past two years, in vivo gene editing, in vivo CAR-T, and mRNA coding TCE technologies have continued to attract multinational pharmaceutical companies (MNCs) to increase their layout. The three have a common direction of innovation: to transfer the production process that used to have to be completed in vitro to the human body's own cells for implementation in vivo.

What is the essence of the competition when large pharmaceutical companies are fully competing for in vivo platform technology and capabilities? In vivo gene editing, in vivo CAR-T, and mRNA coding TCE, which loop do they each complete in vivo?

01. TCE new species

Let's start with the deal between Novartis and AbbVie.

On October 2nd, Novartis acquired exclusive global authorization for ABO2203 from AbbVie with a down payment of $575 million and a milestone payment of up to approximately $7.2 billion, and simultaneously obtained exclusive selection rights for several next-generation assets on the AbbVie RNA platform.

ABO2203 is an mRNA encoded CD19 × CD3 T-cell adapter (TCE) currently undergoing two phase I trials in China.

In fact, CD19 × CD3 TCE itself is not a completely new concept: one end binds to CD19 on B cells, and the other end binds to CD3 on T cells, pulling T cells closer to B cells and killing them.

But ABO2203 is different. It does not directly inject the TCE protein into the human body, but instead loads the mRNA encoding TCE into lipid nanoparticles (LNP), which are briefly produced by human cells after injection.

In other words, the body itself has become a production workshop, and TCE has shifted from "protein drugs produced in vitro" to "pharmacological signals produced in vivo", thus opening up new space in terms of safety, pharmacokinetics, and administration methods.

In the first human B-cell non Hodgkin lymphoma study published by AACR this year, TCE encoded by ABO2203 showed dose-dependent exposure, with a median peak time of about 5.5 days and an average terminal half-life of about 7.9 days.

Compared with the rapid peak and clearance characteristics of protein TCE (P4107) with the same amino acid sequence, the peak concentration of ABO2203 is significantly lower, but the exposure time is longer.

In terms of safety, 9 patients received dose escalation from 3 μ g to 1920 μ g, without reaching the maximum tolerated dose, no grade of CRS or ICANS was observed, and no dose limiting toxicity was observed. Adverse events are limited to grade 1-2 fever, and liver enzyme elevation is only grade 1 with a low incidence rate.
Outside of tumors, autoimmune indications also read signals of "immune reset".

On September 18th, the first human concept validation of ABO2203 treatment for immune thrombocytopenia (ITP) was published in Cell. After treatment, three patients with refractory secondary ITP observed rapid peripheral B cell clearance, platelet recovery, and improvement in disease activity, with follow-up lasting up to six months; The reported adverse events were grade 1 to 2, and no cytokine release syndrome was observed.

From the perspective of administration method, ABO2203 is administered subcutaneously without lymphocyte depletion pretreatment or pre medication. The initial plan is once a week, which can be extended to every two weeks or even every three to four weeks after response.

Although still early clinical data, ABO2203 has advanced a set of technological ideas to human evidence: mRNA can generate TCE in vivo, produce expected cell clearance effects, and exhibit disease improvement signals.

From this, it is not difficult to understand Novartis' "foot in the bill" logic, which is not only the asset ABO2203 itself, but also the new modality of "bringing the TCE factory into the body" behind it.

02. In vivo therapy trend

Placing this transaction in a larger industry context, it will be found that it is in the midst of a continuously heating investment wave: MNCs are continuously competing for in vivo therapy.

Among them, CAR-T therapy in the body is the most intensive direction for large pharmaceutical companies. From 2025 to 2026, the global transaction volume for CAR-T in vivo is expected to reach nearly $20 billion. The participation methods of multinational pharmaceutical companies include both direct acquisitions and technology platform cooperation, and there has been a significant differentiation in delivery pathways.

In March 2025, AstraZeneca took the lead by acquiring Belgian company EsoBiotec for a total consideration of $1 billion, obtaining an in vivo cell therapy platform that utilizes engineered nanobody lentivirus technology and core product ESO-T01 targeting BCMA.

Subsequently, AbbVie announced the acquisition of Capstan for up to $2.1 billion in cash at the time of delivery, advancing the technology of using targeted lipid nanoparticles (LNP) to modify cells in vivo.

Subsequently, Gilead's Kite acquired Interius BioTherapeutics for $350 million, obtaining its integrated lentiviral in vivo CAR-T platform. In addition, Kite has signed a licensing agreement worth up to $1.6 billion with Prikin, completing a dual layout in the field of CAR-T in vivo.

BMS is also not willing to fall behind. In October 2025, BMS acquired Orbital Therapeutics for $1.5 billion in cash, acquiring the core candidate drug OTX-201 and the circular RNA based in vivo CAR-T therapy platform.

Lilly has invested nearly $10 billion in just two months in 2026 to acquire two companies, Orna Therapeutics and Kelonia Therapeutics. The former focuses on the combination of circular RNA platforms and LNP delivery systems, while the latter utilizes the iGPS platform for dual engineering of lentiviral vectors.

Earlier, Eli Lilly acquired Verve for a total consideration of approximately $1.3 billion, including approximately $1 billion in cash ($10.5 per share) and up to $300 million in CVR.

Verve's core pipeline, VERVE-102, is an in vivo base editing therapy targeting PCSK9 gene, which can permanently reduce low-density lipoprotein cholesterol through a single intravenous infusion for atherosclerotic cardiovascular disease.

Lilly positions this transaction as a "one and done" treatment - transforming lifelong chronic management of cardiovascular disease into a one-time gene editing intervention.
Overall, these transactions cover different technologies and pathways, but in terms of transaction volume and density, MNCs have been tirelessly competing for "in vivo editing" platforms and capabilities in the past two years, and are willing to pay large upfront funds for "allowing cells to complete therapeutic tasks in vivo".

In contrast, Novartis' layout in this field is mainly focused on collaborative testing.

In November 2024, Novartis reached a strategic partnership with Vyriad to develop in vivo CAR-T therapy using its active targeting lentiviral vector platform. Vyriad will receive advance payments, milestone payments, and tiered royalties - Novartis' first foray into the in vivo CAR-T field, but no subsequent acquisitions have been made.
In April 2026, faced with industry inquiries triggered by Eli Lilly's two consecutive large acquisitions of CAR-T in vivo, Novartis CEO Vas Narasimhan gave a clear response during the earnings call: the company is still "continuously evaluating" the CAR-T field in vivo, but "there are no ongoing mergers and acquisitions", and if attractive opportunities are seen, they will be seriously considered.

In the end, among several different technological routes such as in vivo gene editing, in vivo CAR-T, and mRNA encoded TCE, Novartis chose to bet on the less crowded mRNA encoded TCE.

03. Differences in Three Paths

Specifically, the attractiveness of in vivo gene editing, in vivo CAR-T, and mRNA encoded TCE lies in transferring the complex work of in vitro GMP workshops to the body, thereby significantly reducing costs. Although all are classified as "in vivo editing" narratives, the three routes have obvious fundamental differences.

In vivo gene editing is the permanent modification of the genome of cells (usually T cells); CAR-T in the body equips T cells with a "radar" (CAR receptor) that recognizes tumors; And mRNA encoding TCE is a temporary production of a "connector" protein that pulls T cells closer to the tumor.

(1) In vivo gene editing: permanent modification of "cell instructions"

This route pursues one-time administration and lifelong benefits. It uses tools such as CRISPR-Cas9 to directly modify the genome of specific cells in the patient's body - possibly by knocking out a pathogenic gene or inserting a therapeutic gene.

Delivery is the biggest bottleneck. The current mainstream strategy is the combination of AAV virus and LNP: LNP delivers CRISPR editing tools, while AAV delivers repair template DNA. The packaging capacity of AAV is limited (about 4.7kb), and large-sized editing tools need to be split into multiple vectors.

The fundamental risk of this route lies in permanence: once an editing error or off target occurs, it is difficult to reverse, so long-term safety monitoring is necessary.

(2) CAR-T in vivo: Equipping T cells with 'radar' in the body

Traditional CAR-T requires extracting the patient's T cells, modifying and expanding them in vitro, and then reinfusing them, which takes 2-4 weeks. The goal of CAR-T in vivo is to bypass the in vitro steps and directly transform T cells into CAR-T in the patient's body.

At present, there are two main technical branches with significant differences:

Virus vector route (lentivirus): It can achieve stable CAR gene integration, and after a single injection, 65% of T cells in non-human primates can express CAR, and B cell clearance lasts for 76 days. But the cost is the risk of random insertion of mutations in the body - at least quality control of cell products can be achieved during in vitro preparation, while in vivo transduction cannot be screened, and the consequences of transduction to non target cells (such as germ cells) are unpredictable.

LNP mRNA pathway: Delivery of CAR encoding mRNA using lipid nanoparticles targeting T cells, mRNA is briefly expressed in the cytoplasm and does not integrate with the genome. But the expression time is very short: the transduction rate of CD3-LNP is about 15% 12 hours after delivery, and drops to around 5% after 60 hours, requiring repeated administration. There is currently no clinical answer on whether repeated administration will produce immunogenicity.

(3) MRNA encoding TCE: does not modify cells, only acts as a 'matchmaker'

The logic of this route is fundamentally different from the previous two. It does not modify any cells, but delivers mRNA into the body (usually the liver or lymphatic organs) using LNP, allowing cells to temporarily produce a bispecific protein. This protein binds to CD3 on the surface of T cells at one end and tumor antigens at the other end, acting as a "bridge" to pull T cells to the edge of the tumor for activation.

The core advantage of mRNA TCE comes from its controllable pharmacokinetics. Traditional protein based TCE requires gradually increasing doses to avoid cytokine release syndrome (CRS), while mRNA TCE achieves progressive drug exposure through a translation process, naturally avoiding the toxicity caused by sudden increases in blood drug concentration.

But the limitations of this route are also evident: the persistence of therapeutic effects depends on continuous administration. The TCE protein translated by mRNA will be gradually cleared, unlike CAR-T which persists in the body. It is more suitable for scenarios that require immune reset but do not require permanent cell modification, such as autoimmune diseases.

Novartis' choice is not so much a trade-off between three routes, but rather a direction that has not yet been heavily bet on after the valuation of the CAR-T track in the body has been pushed up.

MRNA TCE can be discontinued, repeated, and dose adjusted. Novartis uses a reversible, pharmacological "in vivo drug" approach to continue its accumulated cell therapy capabilities in the field of autoimmunity. This may not be the biggest bet, but it could be the one that best matches its own rhythm.

Reference material:
1.128 Jianguo Road, Aibo, finally waiting for the next round
2.What was bought for $575 million in Future Bioinformatics? Re understanding the transaction between Aibo and Novartis
3. Trendy Medicine Bar, difficult to evaluate, what did Novartis buy for $7.2 billion