Physiological Human In Vitro Models to Study and Evaluate Tumor Immunology

Preclinical models such as cell lines, spheroids, and animals are a crucial aspect of modern-day biomedical research. When used correctly, they can provide invaluable insights on safety, efficacy, and mechanistic understanding.

Although the push to reduce animal testing and improve the physiological relevance of in vitro models known as “New Approach Methodologies (NAM)” started more than a decade ago, the FDA’s announcement last year on plans to phase out the requirement for animal testing in immunotherapies like monoclonal antibodies has attracted significant academia and commercial attention–both enthusiasm and concern. This is because while there has been significant progress in developing non-animal models like lab-on-a-chip, organoids, and tissue explants, most published work has not demonstrated functional immune systems in non-animal systems to evaluate immunotherapeutics.¹

The addition of immunity into in vitro models is challenging due to a lack of protocol to maintain tissue immune cell diversity in vitro and differences in immune subtypes between tissues and blood (i.e., peripheral blood mononuclear cells/PBMCs). With rising appreciation of immunity, researchers have found creative ways to incorporate or preserve immune cells in in vitro models.

Chimeric antigen receptor (CAR) T cells have been used to treat hematologic malignancies. At least seven approved CAR T cell products are on the market. However, owing to microenvironmental factors such as a thick stromal barrier and metabolic inhibition, CAR T cells are not as effective to treat solid tumors, necessitating a better model to study tumor-immune interactions. One approach to enable this is tumor-on-a-chip which can be precisely manipulated to mimic different environmental conditions to evaluate the efficacy of adoptive cell therapies.

Liu and co-authors recently developed a tumor-on-a-chip model that contains vasculature that recapitulates the interface of CAR T cells with tumor-associated blood vessels.² The system can easily accommodate spatially defined transplantation of living tumors from xenograft models or human patients to generate vascularized solid tumor constructs and, with the addition of perfusion, be maintained for prolonged periods.

“Tumor-on-a-chip offers new opportunities to reconstruct tumor-immune interactions in vitro and hold exciting potential for efficacy and toxicity evaluation. However, key challenges remain in achieving higher physiological fidelity of these models and in generating meaningful data to inform clinical development of CAR T cell immunotherapies,” says Haijiao Liu, PhD, and author of the paper and postdoctoral researcher at the University of Pennsylvania.

Six steps

Liu’s model is fabricated via six simple steps with reagents that are commonly found in biological labs: a micropatterned insert is added into funnel-shaped wells to block fluidic access;, extracellular matrix (EMC) hydrogel precursor solution containing stromal and endothelial cells are injected into the middle chamber and solidified; more endothelial cells are added to the side chambers to form vasculature networks; tumor is transplanted into the vascular bed by removing the insert before sealing the access with acellular ECM hydrogel; and perfusion can be performed with embedded tumor explants and surrounding blood vessels.

The authors first used their model to emulate three critical steps of tumor-directed CAR T cell therapy, including extravasation and infiltration, recognition of tumor-associated antigens, and anti-tumor immunity and persistence. As a proof-of-concept demonstration, the team used A549 human lung adenocarcinoma cells engineered to overexpress mesothelin and mesothelin-targeted CAR T cells from three donors. The team found that endothelial expression of intercellular adhesion molecule-1 (ICAM-1) was substantially higher in the tumor-associated vessels of meso-tumors, which was consistent with previous literature reports.

Using human tumor explants, the authors further discovered that there was active crosstalk between CAR T cells and endothelial cells. When this interaction was blocked with anti-CD38 antibody, there was significantly reduced CAR T trafficking into the tumors which limited tumor killing. Overall, the data suggest that this system can work to model human tumor-immune interactions on a chip at the cellular level and identify novel molecular targets to enhance CAR T therapy efficacy when used with advanced omics studies.

Liu adds that their system faithfully recapitulates key features of the tumor microstructure, including cancer cells, vasculature, and fibroblasts. His team’s system enabled direct visualization in real time of how lung cancer-targeting CAR T cells adhere to blood vessels, extravasate, and infiltrate tumors, processes that have been difficult to observe in conventional models.

Their tumor-on-a-chip also functioned as a powerful preclinical data-generation platform, enabling multi-omics analyses that produced as or more informative, predictive, and translatable to human insights than those produced by conventional in vitro or animal studies. Their model also revealed targetable signaling pathways that limit endothelial recruitment of CAR T cells.

Lymphoma organoid

An immune organoid model that contained digested immune tissues before mixing with ECM and dermal fibroblast cell line to study human immunity was previously described in GEN. Taking a similar approach but with PBMCs from lymphoma patients, the team generated cancer organoids to study how this disease makes patients more vulnerable to infections and autoimmune diseases.³

The PBMCs came from patients who were undergoing remission follow chemotherapy. The team found that compared to PBMC organoids derived from healthy donors, the lymphoma organoids had fewer germinal center B cells and plasma cells after treatment with Fluzone vaccine. The germinal center B cells are a crucial cell type for recognizing antigens and to mount immune responses. This was later found to possibly be attributed to the failure of B cells to form germinal centers even with the addition of CXCL12 chemokine.

The authors hypothesized that post-chemotherapy B cells might lose ability to be spatially partitioned in immune tissues, resulting in compromised immunity after chemotherapy.

Unlike other immune organoid models, this system effectively made use of accessible PBMCs with ECM engineering and stromal cells to produce a synthetic model that captured immune diversity and maintained immune cell viability. It represents a major step using immune-cancer organoids as a system for scalable drug screening.

“Complex in vitro models, as described by the authors, are in urgent need to complement animal studies, which itself is facing scrutiny in terms of relevancy and quality of translatable data,” says Jeremey Teo, PhD, associate professor at New York University, Abu Dhabi.

Teo adds that one can and should integrate relevant cancer-associated stromal cells, engineered tissue surrogates mimicking TME microarchitecture, and lymphoid compartments. Increased intricacy must be balanced by ease of reading and interpreting the biological outcomes using current instrumentation. One such recent publication from his lab describes the formation of tertiary lymphoid organoids that incorporated stromal-immune interactions to reflect the role of stromal cells in shaping tissue immunity.⁴

Preserving tumor explants

Patient-derived tumor explants (PDTE) are gaining traction as a model to study tumor immunity because they preserve spatial heterogeneity of tumors, cell diversity, and ECM-cell interactions. However, due to the need for preparation steps like slicing and changes in the metabolic environment, most PDTEs can only stay viable for two to four days, which limit their use for drug screening such as for drugs that require repeated doses over >4 days.

Peritoneal metastases (PM) arise from the dissemination and outgrowth of cancer cells from abdominal and pelvic tumors to the peritoneal cavity. Wu and colleagues developed a customizable hyaluronic acid hydrogel to embed PM PDTE and study the factors influencing explant viability before evaluating it for chemotherapeutic efficacy.⁵

The scientists relied on hyaluronic acid hydrogel because it is tunable in material properties, biocompatible, and can be rapidly crosslinked to enable 3D embedding of PDTE. Unlike Matrigel and inert polytetrafluoroethylene which led to significant shrinkage of PDTE after culturing, hyaluronic acid hydrogel preserved >72% of the slice area and >60% of the cells were viable even at day 12.

The team found that when cultured in a 3D environment, PDTE always survived better than in a 2D environment, even when both environments had hyaluronic acid hydrogel. Using bulk RNA sequencing, they discovered that PDTE cultured in 3D hydrogel had lower expression of genes associated with inflammation, apoptosis, and necrotic cell death. Working with a variety of tissue staining and imaging techniques, the team discovered that ECM structural modifications, especially the loss of collagen structures, led to tissue contraction and cell death.

When PDTE was cultured with inhibitors that disrupted myosin II, myofibroblasts were able to produce collagen and maintained better slice viability and overall cell diversity, including T cells, B cells, macrophages, and mast cells, compared to the original tumors.

“This research was motivated by the need for more improved tumor models that capture the full extent of tumor complexity and heterogeneity. Besides cancer cells, there are also non-cancer supporting cells that foster tumor progression and drug resistance; these are currently poorly modeled after in existing tumor models,” notes Eliza Fong, PhD, senior author of this study and an assistant professor at the National University of Singapore.

“We found that it is possible to extend the lifespan of patient-derived tumor fragments outside the body using engineered biomaterials, keeping them alive long enough for drug evaluation. Given that these models fully preserve the immune landscape within the tumor intact, these models will likely be leveraged for the evaluation of immunotherapies (immune checkpoint inhibitors, T-cell engagers, etc.) in an all-human context.

Treating tumors with immunotherapy is highly promising but solid tumors have evolved mechanisms such as creating thick stromal barriers, downregulation of cancer-associated antigens, metabolic inhibition, and recruitment of tumor-associated cell types to limit the efficacy of immunotherapies. Predictive in vitro human models play a crucial role in overcoming limitations associated with simpler models and animals to understand tumor-immune interactions, mechanism studies, and drug screening.

Data generated from tumor-on-a-chip, organoids, and tissue explants can also be used to build virtual tumor models for digital twin studies. This will fulfil the promise of NAM to positively advance tumor immunology research.

References

1. Wang D, Villenave R, Stokar-Regenscheit N, Clevers H. Human organoids as 3D in vitro platforms for drug discovery: opportunities and challenges. Nat Rev Drug Discov.Nature Research. Preprint posted online 2025. doi:10.1038/s41573-025-01317-y
2. Liu H, Noguera-Ortega E, Dong X, et al. A tumor-on-a-chip for in vitro study of CAR-T cell immunotherapy in solid tumors. Nat Biotechnol. Published online 2025. doi:10.1038/s41587-025-02845-z
3. Zhong Z, Quiñones-Pérez M, Dai Z, et al. Human immune organoids to decode B cell response in healthy donors and patients with lymphoma. Nat Mater. Published online 2024. doi:10.1038/s41563-024-02037-1
4. ElGindi M, Karaman S, Teo J. ⁠Engineering Adaptive Immunity in 3D: A Patient-Specific Lymphoid Model Using Stromal Networks and Peripheral Blood Mononuclear Cells. Advanced Science. Published online 2025. doi:10.1002/advs.202513245
5. Wu KZ, Ding RH, Zhao Z, et al. Hydrogel-Mediated Preservation of Live Tumor Explants for Drug Development in Peritoneal Metastases. Advanced Materials. 2025;37(33). doi:10.1002/adma.202418647