Cancer immunology assays in drug discovery

What these assays are for

Cancer immunotherapy, including checkpoint inhibitors and engineered cell therapies, works by acting on immune cells rather than on the tumor directly. That changes what a screening assay has to measure. A conventional cytotoxicity readout on a tumor cell line will not detect an agent whose effect depends on a T cell being present, so the assays used in these programs are built around an effector population and a target population together. The question each assay answers is narrow: whether the candidate increases killing of the target, whether it engages a specific receptor, or whether it releases a brake on immune activation.

The main assay types

Cytotoxicity (killing) assays

A killing assay measures how effectively effector cells such as T cells or natural killer cells destroy target tumor cells. Several formats are in use, and a published comparison of them sets out where each is appropriate: chromium release, bioluminescence, impedance-based real-time monitoring, and flow cytometry.^[S2] They differ in what they actually detect. Enzyme release assays, including the lactate dehydrogenase format, have a known limitation in this setting: effector cells that die during the assay release the same enzyme as the target cells, so the signal is not specific to target killing.^[S2] Impedance measurement gives a label-free, continuous readout of adherent target cells over hours, while flow cytometry resolves which population died. The choice follows from the mechanism being tested, not from which format is newest.

Antibody-dependent and complement-dependent killing

These assays are used mainly for antibody therapeutics. Both mechanisms describe how an antibody bound to a target cell leads to that cell being destroyed, but by different routes. In antibody-dependent cell-mediated cytotoxicity, effector cells such as natural killer cells and macrophages recognize the bound antibody through their Fc receptors and kill the target. In complement-dependent cytotoxicity, complement proteins in serum bind the antibody and lyse the cell directly. An antibody may work through one, both, or neither, so the two are measured separately.

Reporter bioassays are a common alternative to measuring killing directly. Instead of counting dead cells, an engineered effector cell carries a reporter gene, often luciferase, downstream of the signaling pathway of interest; light output then reports pathway activation. Published protocols describe this format for antibody-dependent cytotoxicity.^[S3] The practical advantages are a wide dynamic range, low variability and no requirement for donor cells, which makes the assay suitable for release testing and for comparing lots. The trade-off is that a reporter measures receptor engagement and signaling, not the killing itself.

Cytokine production and secretion

Activated immune cells release cytokines, and measuring them describes the character of the response rather than its magnitude alone. Three formats are commonly used. The enzyme-linked immunosorbent assay quantifies a single analyte in a sample, and the Food and Drug Administration names it, alongside quantitative flow cytometry, as an example of an immunochemical analytical method in its guidance on potency testing for cell and gene therapy products.^[S1] The enzyme-linked immunospot assay detects secretion at the level of individual cells, which answers a different question: how many cells are responding, rather than how much is present overall. Multiplex assays measure many analytes in one sample, at the cost of more complex validation. Cytokine release is also a safety signal in immunotherapy, so a rise in these readouts is not automatically a favorable result.

Checkpoint binding and function

Checkpoint molecules act as brakes on immune activation, and an antibody intended to release that brake is characterized in two steps. A binding assay measures how well the candidate blocks the interaction between the receptor and its ligand. A cell-based reporter assay then tests whether blocking that interaction produces the intended signaling change in a cell, which is the step that reflects the proposed mechanism of action. Published work describing such a bioassay for a bispecific antibody against two checkpoint targets illustrates the design.^[S4] Binding affinity alone does not establish functional activity, which is why both are usually run.

More complex culture systems

Organoids and spheroids with immune cells

A monolayer of tumor cells cannot represent the tumor microenvironment, which includes stromal cells, vasculature and suppressive immune populations as well as the tumor itself. Three-dimensional cultures, including patient-derived tumor organoids and spheroids co-cultured with immune cells, allow some of those elements to be included, and they allow the physical problem of immune cell infiltration into a solid mass to be observed at all. What this adds is the ability to measure things a flat co-culture cannot show. It does not follow that results in these systems predict clinical outcomes more accurately, and this article makes no such claim.

Image analysis and high-content screening

High-content screening combines automated microscopy with image analysis, including machine learning approaches, so that many features can be extracted from each well rather than a single number. In three-dimensional immune co-cultures this allows a time course to be followed in the same wells, and it allows changes in cell morphology and apoptosis to be quantified rather than inferred from a bulk viability reading. The expectation is that this improves the efficiency of working through a compound set; no figures for that improvement are cited here, because none could be substantiated.

Running these assays in house or placing them outside

What building the capability involves

Establishing these assays internally involves several problems that are specific to working with primary immune cells. Peripheral blood mononuclear cells have to be sourced and prepared consistently. Donor-to-donor variation is substantial and has to be managed by design, typically by running several donors and treating donor as a factor rather than averaging it away. Co-culture conditions, including the effector-to-target ratio and the assay duration, have to be optimized for each target. Instrumentation for real-time cytotoxicity monitoring or three-dimensional imaging is a further commitment, in maintenance and scheduled use as much as in purchase. No cost figures are given here because none could be substantiated.

What to settle before outsourcing

Placing the work with a laboratory that already runs these assays avoids the setup described above. Whether it is the right choice depends on the answers to a few specific questions, which are worth settling in the statement of work rather than afterwards: how many donors will be used and whether the same donors will cover the whole study; how the effector-to-target ratio and assay duration were chosen for this target; which positive and negative controls appear on every plate; how a result is called positive; whether the raw data, including images and flow cytometry files, will be transferred or only the summary; and whether any of the work will be subcontracted.

Where this is heading

Work is moving toward culture systems that include more of the tumor microenvironment than tumor cells and effector cells alone, adding vasculature and suppressive populations such as regulatory T cells and myeloid-derived suppressor cells. The reason is mechanistic rather than fashionable: an agent whose intended effect is to relieve immune suppression cannot be characterized in a system that contains no suppression. Whether such assays become the usual choice is not something this article predicts.

Patient-derived tumor organoids are also being investigated in the context of individualized treatment selection. That is a clinical research question and is outside the scope of this site, which covers nonclinical work. For a developer, the nonclinical use of a panel of patient-derived organoids is different and more immediate: it gives an early indication of whether an effect is confined to particular tumor genotypes, before an animal study is designed.