February 1, 2018 (Vol. 38, No. 3)

How to Gauge Your Investigational CAR-T Therapy

Chimeric antigen receptor T-cell (CAR-T) therapies have shown great promise for the treatment of certain types of hematological cancers, with two products approved to date by the FDA with more in the pipeline.

With this great potential comes a challenge—no comprehensive industry-wide safety-testing procedures have been adopted for these innovative therapies. Safety testing for CAR-T products is not standardized, and the level of testing required may vary depending on the specificity of the CAR and other features. Unlike safety testing for small-molecule and biologic therapies, a one-size-fits-all testing approach will be more difficult to establish and implement. Therefore, developers of CAR-T therapies will need to consider a long list of variables before embarking on safety testing.

Prior to Conducting a Preclinical Animal Study

In early-stage studies, it is important to fully understand how the CAR-T therapy has been engineered and how this might impact both efficacy and safety. The following questions should be addressed before embarking on animal studies:

What is the nature of the target being recognized—mouse or human—and where is the target expressed, what is the target density, and how is it regulated?
It is most important to know whether the target is ever expressed on normal cells. Ideally, the target molecule should be expressed only on the tumor cell and should not cross-react with epitopes expressed on healthy tissue. Screening of normal tissue arrays for target expression by immunohistochemistry and/or PCR should be performed.

How strongly is the chimeric receptor expressed on the surface of the CAR-T?
Are expression levels low, medium, or high? This is important to know because the level of expression could influence the strength of the response and the likelihood of toxicity. It is also important to consider whether the level of expression of the CAR can be controlled, and whether the CAR is engineered to integrate randomly in the genome or inserts in a specific location, as this can impact function.

What costimulatory and signaling domains are expressed by the CAR-T?
This is important to know, because increased signal strength and higher T-cell activation could lead to a wider spectrum of toxicities or improved efficacy.

Has the CAR-T product been engineered to express additional molecules that could affect the safety profile (e.g., cytokines), or molecules that modulate its activity (e.g., deletion of immunoregulatory molecules such as PD-1)?
Looking at how a CAR-T construct is specialized is crucial.

Is the CAR-T drug engineered in such a way that they can be turned off or killed?
These features are important because the ability to control CAR T-cell activity or to induce suicide enhances the safety profile of the product.

Has the CAR-T therapy been engineered to eliminate expression of the endogenous T-cell receptor (TCR) and/or major histocompatibility complex (MHC)?
This is important because CAR-T constructs lacking TCR expression would be less likely to induce graft versus host disease; CARs lacking MHC would not be subject to rejection. These modifications have been proposed as a way to develop off-the-shelf CAR-T therapies that could be used in any patient, irrespective of human leukocyte antigen (HLA) status.

Is the cellular product defined in composition and phenotype?
That is, do we know whether the product is comprised of CD8, CD4, or a mixture of both cell types? What is the optimal ratio? Have the CAR-T cells been cultured in vitro to induce differentiation to a specific phenotype (e.g., central memory, effector memory)?

What in vitro functional characterization has been performed prior to animal testing?
Do we know anything about the cytotoxic function of the product’s CD8 component, its potency, or the role of the CD4 component and its cytokine production?

In Vivo Studies

Animal studies should attempt to further explore issues such as pharmacokinetic/pharmacodynamics (PK/PD), safety, and efficacy. Study designs should consider the following factors:

As with any other form of therapy, identifying the optimal/maximum dose is essential. The optimal dose will vary for different CAR-T programs, but the objective is always to identify the dose which results in greatest efficacy with minimal toxicity. If the optimum has not been estimated by in vitro studies, we recommend testing at least three different doses.

For tumor studies, it is important to identify when to begin treatment with CAR-T after implanting tumor cells. The timing will vary depending on the growth characteristics of the tumor model. A higher antigen load (i.e., larger tumor) may result in more potential for toxicity, but may be of greater clinical relevance, as well.

Persistence and Expansion
How long does the CAR-T candidate persist in the animal? Can the investigational therapy be tracked easily in blood? Many investigational CAR-T therapies are engineered to express a tag or marker that allows detection by flow cytometry. Multiple longitudinal time points should be identified to perform sampling and detection of CAR-T products. Flow analysis will provide information about the relative expansion, activation, and persistence of the cells.

Cytokines and Biomarkers
Cytokine release syndrome is one of the most serious clinical side effects associated with administration of CAR-T therapies. Although mouse models may not reflect the full spectrum of cytokine-mediated events seen in humans, it is important to analyze serum cytokine levels at multiple time points in any preclinical CAR-T study. Analysis should be performed at multiple time points. Additional biomarker analyses can also be performed on the samples.

Blood Pressure and Body Temperature Measurements
Changes in blood pressure and body temperature may mimic some clinical manifestations seen in humans. The most sensitive methods require implantation of telemetry monitoring devices.

Weight Loss and Detailed Clinical Observations
These measurements provide important information about treatment tolerability and the health status of the animals.

Organs should be collected at the end of the study and subjected to full histopathological analysis.

Whenever possible, include a group that receives a scrambled (negative control) or non-signaling CAR. 

Model Selection
The most useful models for CAR-T studies include:

  • Engraftment of CAR-T in tumor-bearing immunocompromised mice (e.g., NSG, NCG mice). This model is useful for testing human CAR-T therapies and can be used to assess activity against human tumor cell lines (CDX) or patient-derived tumors (PDX).
  • Engraftment of CAR-T in humanized mice immunocompromised mice humanized with CD34 HSC or PBMCs. This model allows for the assessment of CAR-T activity in the presence of human immune cells, which increases its clinical relevance. These models are expensive, however, and may be impacted by the development of graft-versus-host disease and the effect of HLA.
  • Engraftment of murine CAR-T in a fully immune-competent host. CAR-T therapy safety can be evaluated in mice bearing syngeneic murine tumors or in nontumor-bearing mice.

For CAR-T studies using disseminated tumor models (e.g., Raji-B, Nalm-6) consideration should be given to the use of luciferase-transduced tumor cells and bioimaging to monitor tumor growth. It is also feasible to perform dual imaging of CAR-T cells and tumor cells. This allows for sensitive tracking of CAR-T cell trafficking.


Because there is no standard approach to safety assessment, and clinical trials of CAR-T therapies are likely to be even more expensive than standard trials, any knowledge obtainable from early stage studies is extremely valuable. Developers of CAR-T products should consider study designs that broadly address both safety and efficacy endpoints to reduce development costs, shorten time to approval, and mitigate risk when the product moves into human testing.

David P. Harris, Ph.D, is research director, oncology discovery at Charles River Laboratories.

Previous articleDeveloping a Diagnostic for Alzheimer’s Disease
Next articleRPA Kits