The Conference
The American Society of Gene and Cell Therapy annual meeting wrapped in Boston the week of May 12, 2026. Endpoints' on-the-ground coverage described in vivo CAR-T as one of the dominant themes, while noting that much of the field remains preclinical. Tessera reported preclinical in vivo Gene Writing data in sickle cell disease, including up to 85% correction of the sickle-cell mutation in human HSCs in a mouse model, alongside separate progress in in vivo CAR-T applications. Encoded presented clinical Phase 1/2 POLARIS data on its ETX101 gene regulation therapy for SCN1A-positive Dravet syndrome, with seizure reductions through 52 weeks after a single dose — adjacent to, but not the same modality as, in vivo CAR-T.
For regulatory teams, the hard problem is not simply whether the biology works. It is how to write the first generation of IND-enabling packages for modalities where the precedent base is thin, fragmented, and borrowed from adjacent categories.
Why In Vivo Changes the Writing
The mature ex vivo CAR-T category has a richer regulatory precedent base. The public record around Kymriah, Yescarta, Tecartus, Breyanzi, Carvykti, and related products gives writers a known vocabulary: cell collection, viral vector manufacture, transduction, expansion, cryopreservation, release testing, potency, comparability, and patient-specific chain of identity. A writer drafting Module 3 for an ex vivo CAR-T program in 2026 has a much richer public and regulatory precedent base than a writer drafting for in vivo CAR-T.
In vivo cell engineering shifts the center of gravity. The product is no longer an engineered cell administered back to the patient. It is a delivery-and-payload system — for example, an LNP, viral vector, mRNA, DNA, or editing construct — intended to engineer cells inside the body. That shifts the writing burden from "how was the cell product manufactured and released?" to "how does this construct reach the intended cells, express or edit the intended payload, avoid unintended cells, and produce a predictable biological effect?"
That is why the CMC and nonclinical narratives become inseparable in this category. Biodistribution, target-cell specificity, potency, persistence, immune activation, off-target delivery, and — where relevant — off-target editing are not separate technical issues. They are one connected argument about whether the administered construct can safely and reproducibly create the intended cellular phenotype in vivo.
What the IND Writer Inherits
Three sections of the IND will sit on a thinner and more fragmented precedent base than writers are used to.
The biodistribution and target-cell specificity narrative. Existing gene therapy guidance, including ICH S12, gives a framework for biodistribution, persistence, and tissue exposure. But in vivo CAR-T and related in vivo cell-engineering platforms may require that framework to be adapted to cell-type specificity, immune-cell targeting, and downstream cellular phenotype. The writing has to position the candidate within the existing framework while acknowledging where the framework needs extension. This is a delicate posture; over-claiming framework fit invites IRs, but framing the candidate as fully novel forfeits the precedent corpus the writer could cite.
The off-target activity narrative. For editing-capable systems, this includes off-target editing. For transient-expression or non-integrating systems, the more relevant concerns may be off-target delivery, off-target expression, immune activation, persistence, and reversibility. The writer has to avoid importing the wrong risk vocabulary from adjacent modalities. A narrative that treats a transient mRNA construct as if it were a CRISPR base editor is a writer's error that a reviewer will catch in the first read.
The potency and release-testing narrative. Release testing measures the administered delivery-and-payload system. The therapeutically relevant cellular phenotype appears later inside the patient. The IND has to make the chain of reasoning explicit: which product attributes are being measured, how those attributes predict target-cell engineering, and why that engineered cellular phenotype supports the clinical hypothesis. This section is where the FDA's CMC information for human gene therapy IND applications gets adapted, but adaptation is not the same as inheritance — every assay, every specification, and every justification has to be written for the specific platform.
The Cross-Sponsor Convergence Problem
The teams filing first will help set the precedent for everyone else. The agency reads these submissions in sequence. The first wave of INDs will likely influence how reviewers think about recurring questions such as biodistribution scope, target-cell specificity, persistence, potency, and off-target activity.
If the first filings use disparate conventions for the biodistribution narrative, the agency's read stays unsettled for years. If they converge on a similar narrative structure — using similar headings, similar precedent citations, similar evidence pyramids — the agency develops a stable reading frame much faster, and every subsequent sponsor benefits.
This convergence does not happen automatically. It happens because the writers on the first programs read each other's published documents and the FDA's cellular and gene therapy guidance corpus, and pattern-match on what worked. In categories with established precedent corpora, the convergence is implicit in the existing FDA guidance. In categories where the corpus is borrowed from adjacent modalities, the convergence has to be manufactured by the writers themselves.
In practice, the first few sponsors will help establish the vocabulary that later FDA guidance may formalize. Whether they realize it or not.
What This Means for Today's Programs
For a sponsor whose in vivo cell-engineering program is preclinical today, four practical takeaways.
Start drafting the CMC narrative earlier than feels necessary. The chain of reasoning between delivery-and-payload attributes and cellular outcome takes longer to articulate than the team expects. Drafting starts when the platform's pharmacology and biodistribution picture is stabilizing, not when the IND meeting is scheduled.
Read adjacent precedent carefully, and label what is borrowed. AAV gene therapy filings are not the same modality, but they establish framework language for biodistribution, release testing, and potency assays that is partially transferable. Ex vivo CAR-T filings supply vocabulary on cellular phenotype, potency-to-clinical-outcome bridging, and immunogenicity. The writer who has read those filings starts with a better vocabulary than the writer starting from first principles — but the writer also has to be explicit about what is being borrowed and where the borrowing breaks.
Plan a pre-IND meeting earlier in the program calendar than usual. The agency's interpretive lens on in vivo cell engineering is being formed in real time. A pre-IND meeting is the cheapest way to learn how the reviewer is currently thinking. Sponsors that wait until twelve months before IND are getting the reviewer's view from twelve months ago.
Treat the writing function as a category-shaping investment, not a filing cost. The first sponsor to file a clean, well-structured in vivo cell-engineering IND establishes vocabulary that the next sponsors will inherit. That investment compounds for the program and for the platform.
The Larger Pattern
ASGCT 2026 showed that in vivo cell engineering is moving fast. But the regulatory writing challenge is not solved by better conference data. It is solved by building a coherent bridge between product attributes, biodistribution, target-cell specificity, potency, and clinical hypothesis.
In established categories, writers inherit the corpus. In emerging categories, writers help build it. The first wave of in vivo CAR-T and adjacent in vivo cell-engineering INDs will not just describe a new modality. They will help define the vocabulary regulators and sponsors use to evaluate it.