In Vivo's Biggest Threat — Comparison To Old Models

By Arnaud Deladeriere, Ph.D., Cell&Gene Consulting Inc.

In vivo CAR-T therapy represents a distinct paradigm. The industry must start treating it that way and move beyond high-level comparisons to ex vivo models if we will ever treat disease with it.

This was the premise underlying the latest Cell&Gene Foundry session, which focused on biology, analytics, clinical translation, and commercial implications. We set out to identify where the promise of in vivo aligns with our current understanding, and where meaningful uncertainties remain.

The outcome was a balanced assessment. In vivo CAR-T is neither a straightforward evolution of ex vivo CAR-T nor a purely manufacturing-driven innovation. A central theme of the discussion was control, specifically how control is redistributed when CAR engineering moves from an ex vivo environment into the patient.

Ex vivo CAR-T development has been shaped by the ability to define and monitor key variables, including:

• starting T-cell population and subset composition
• activation and expansion conditions
• transduction efficiency and copy number
• gene insertion strategy and expression control
• selection of a more purified population following engineering (e.g., allogeneic approaches)
• dose defined at the cellular level.

Over time, the field has learned that these parameters meaningfully influence safety, durability, and clinical response.

In vivo CAR-T necessarily relinquishes many of these controls. Once a vector is administered systemically, the developer no longer directly controls:

• which cells are modified
• how many copies are delivered per cell
• how heterogeneous the modified population becomes
• how expansion unfolds across tissues.

Even with advances in targeting ligands or engineered tropism, the resolution remains limited compared to ex vivo manipulation.

This loss of control introduces several practical considerations:

• off-target transduction, including modification of non-T-cell populations such as hematopoietic progenitors
• subset ambiguity, with CAR expression distributed across heterogeneous T-cell populations
• dose opacity, where vector dose does not translate cleanly into a defined cellular dose.

The discussion emphasized that these are design constraints rather than failures. In vivo CAR-T is not simply ex vivo CAR-T without a manufacturing step. It is a different modality that relies more heavily on biological systems to regulate outcomes that are tightly managed ex vivo.

Biological Reality: Targeting 'T Cells' Is Not Sufficient

The discussion repeatedly returned to the biological complexity of T cells. While in vivo CAR-T approaches often describe T cells as a single target population, therapeutic outcomes are strongly influenced by T-cell heterogeneity.

Key biological factors include:

• differentiation and memory state
• metabolic fitness
• exhaustion and prior antigen exposure
• cytokine responsiveness.

Ex vivo CAR-T programs have learned that modifying the wrong subsets can limit persistence or increase toxicity. In vivo approaches currently lack the ability to reliably select or enrich for favorable subsets.

This limitation is particularly relevant in patient populations with extensive treatment histories. In vivo CAR-T relies on endogenous T cells that may already be dysfunctional, exhausted, or skewed by prior therapies. In this respect, in vivo CAR-T shares biological constraints with autologous ex vivo CAR-T, without the opportunity to optimize starting material.

In autoimmune indications, the challenge is further compounded. Modifying T cells in an already dysregulated immune system without precise control raises the possibility of unintended immune effects. The discussion highlighted that immune modulation in vivo requires a clear understanding of which populations are being altered, and to what extent.

Vector Choice And Durability: A Clinical, Not Technical, Decision

Vector selection was framed as a defining element of the therapy, rather than a delivery optimization.

Lentiviral approaches

Lentiviral vectors offer durable expression through genomic integration and benefit from natural tropism for T cells. This makes them attractive for oncology indications where sustained CAR expression may be required.

However, durability introduces trade-offs:

• random integration and variable copy number
• difficulty assessing insertional risk when transduction occurs systemically
• limited ability to control expression levels in vivo.

In ex vivo settings, these risks are mitigated through extensive characterization. In vivo, they are harder to quantify and manage.

LNP and mRNA-based approaches

LNP-based delivery, particularly with mRNA payloads, offers transient expression and avoids integration-related risks. At the same time, transience introduces different challenges:

• inefficient endosomal escape, reducing functional delivery
• patient-specific serum protein interactions that alter biodistribution
• variability in targeting and expression between patients.

Transient expression may be appropriate for indications such as autoimmune disease, where modulation rather than elimination is the goal. In oncology, the durability required to achieve sustained control remains an open question.

The discussion converged on a key point: durability requirements should be driven by indication and biology, not by delivery convenience or platform familiarity.

Lymphodepletion And Immune Context

One of the frequently cited advantages of in vivo CAR-T is the potential to avoid lymphodepletion. While appealing from a patient experience perspective, the discussion highlighted several biological implications.

In ex vivo CAR-T, lymphodepletion serves multiple functions:

• creating immunological space for CAR-T expansion
• reducing competition for cytokines
• dampening host immune clearance mechanisms.

Removing lymphodepletion without replacing these functions may limit expansion and persistence.

There was also discussion around whether lymphodepletion may be necessary to protect the therapy itself, particularly in preventing rapid immune-mediated clearance of vectors or newly modified cells.

The group agreed that avoiding lymphodepletion remains an attractive goal, but its feasibility is likely indication- and modality-dependent rather than universal.

Analytical Challenges: Defining And Measuring The Active Ingredient

One of the most complex issues discussed was analytical characterization.

In vivo CAR-T challenges conventional regulatory logic because the administered product is not the active therapeutic agent. The vector acts as a precursor, while the true active ingredient is the modified T cell generated inside the patient.

This raises several questions:

• How is batch consistency defined?
• What does potency mean in this context?
• How is control demonstrated when key biological steps occur after administration?

Potency assays in cell therapy have increasingly been expected to predict clinical efficacy, an expectation that is already difficult to meet ex vivo. In vivo CAR-T further complicates this approach.

The discussion suggested that a more pragmatic framework may be required, focusing on:

• vector identity, integrity, and reproducibility
• transduction capability in representative systems
• basic functional activity thresholds.

Under this model, clinical efficacy would be established through clinical data rather than release testing. This represents a shift in emphasis rather than a lowering of standards.

Regulatory Uncertainty And The Role Of Precedent

Regulatory uncertainty was viewed as structural rather than temporary. Existing frameworks were not designed for therapies where the active ingredient is generated in vivo.

Unlike traditional gene therapies, where the vector itself is the drug, in vivo CAR-T occupies a hybrid space between gene therapy and cell therapy. The product itself (the vector) is a gene therapy, but the “active ingredient” is a T cell, engineered by use of the infused vector. Sponsors therefore carry a significant burden of scientific justification for analytical strategies, release criteria, and long-term safety monitoring.

Early clinical programs may benefit from flexibility, but later programs are likely to be benchmarked against both ex vivo CAR-T and emerging in vivo precedents. In this context, early development decisions may have lasting influence on regulatory expectations for the field.

Commercial Considerations: Cost, Price, And Access

The discussion also addressed the assumption that lower manufacturing complexity would potentially lead to lower prices.

While in vivo approaches may reduce certain operational costs, pricing in advanced therapies historically has been driven by:

• clinical value and durability
• competitive landscape
• cumulative R&D investment.

Examples from allogeneic cell therapies demonstrate that reduced manufacturing cost does not automatically translate into lower price tags. Without changes to reimbursement models, in vivo delivery alone is unlikely to reset pricing dynamics.

Scale remains a critical variable. Manufacturing efficiencies are most impactful in large indications. For rare diseases or highly personalized applications, in vivo delivery does little to escape the economics of low-volume development.

Five Key Takeaways

Closing Perspective

In vivo CAR-T represents a meaningful direction for the field and addresses real limitations of current ex vivo approaches. At the same time, it introduces a different set of constraints that must be addressed deliberately.

The discussion emphasized that progress will depend on aligning enthusiasm with biological understanding, analytical rigor, and realistic expectations. Ex vivo CAR-T has benefited from more than a decade of iterative learning. In vivo approaches are earlier in that journey.

Their long-term success will depend on how thoughtfully the field navigates the trade-offs between control, biology, scalability, and access. In the end, there will likely not be an either/or outcome for these two modalities. Instead, they are likely to be deployed strategically depending on indication, condition of the patient, and biological outcome needed to control the specific elements of the disease.

About The Author:

Arnaud Deladeriere, Ph.D., is principal consultant at Cell&Gene Consulting Inc. Previously, he was head of MSAT and Manufacturing at Triumvira Immunologics, and before that, manufacturing manager at C3i. He received his Ph.D. in biochemistry from the University of Cambridge.