Next-Generation Cancer Models for Drug Testing: Recent Advances in Immunocompetent Microphysiological Systems

Abstract

The success of checkpoint inhibitors in improving cancer patient survival has demonstrated the therapeutic potential of immunotherapies. This advancement has reshaped oncology treatment and driven interest in harnessing immune modulation for a wider range of diseases. However, developing drugs that modulate immune activity presents unique challenges. A major limitation in preclinical research is the inefficiency of testing human-specific immune targets in animal models, which often fail to translate to clinical outcomes. Additionally, conventional in vitro systems lack immune reactivity due to their static and monocellular nature, limiting their predictive value. Advanced in vitro models can bridge this gap by offering increasingly relevant human physiology for testing drug efficacy and safety, along with absorption, distribution, metabolism, and excretion (ADME). In particular, immune-competent spheroids, organoids, and organs-on-a-chip (OoC) have emerged as promising tools. Although still in their infancy, these microphysiological systems (MPSs) have demonstrated the feasibility of replicating immune responses ex vivo, providing a new avenue for studying immune-targeting drugs with higher translational potential. In this review, we explore recent advances in immune-competent organoid and OoC models, highlighting their capabilities and limitations. We provide a perspective on their applications for cancer drug testing, discussing how these systems could refine preclinical immuno-oncology research and accelerate the development of next-generation immunotherapies.

1. Introduction

Cancer remains a leading cause of mortality worldwide, requiring continuous innovation in therapeutic strategies and model systems for drug development [1]. According to the World Health Organization, cancer accounted for nearly 10 million deaths globally in 2022, and its incidence continues to rise with aging populations and evolving risk factors [2]. While traditional treatments such as chemotherapy and radiation therapy have long been the foundation of oncology, the clinical focus is increasingly shifting towards immunotherapeutic approaches. These therapies harness the body’s immune system to detect and destroy tumor cells, offering more targeted and, in some cases, long-lasting responses compared to conventional therapies [3,4].

Among immunotherapies, immune checkpoint inhibitors (ICIs) have gained extensive attention since their initial approval in 2011 [5]. These agents, which block regulatory pathways such as PD-1/PD-L1 and CTLA-4, have led to significant clinical successes in several cancers, including melanoma, non-small cell lung cancer, and renal cell carcinoma [6]. As of March 2025, eleven ICIs have been approved for cancer treatment, reflecting their rapid expansion in clinical use [7,8]. Additionally, chimeric antigen receptor (CAR) T-cell therapies have shown remarkable success, particularly in treating certain hematologic cancers [9]. There is also a renewed interest in cancer vaccines and other modalities aimed at boosting immune recognition of tumors [10]. Collectively, these advances highlight the growing importance of immunotherapy in modern oncology and its expected dominance in future cancer treatment strategies.

While significant progress has been made in the development of anti-cancer therapies, researchers continue to face several challenges in preclinical testing. Alarmingly, fewer than 11% of anticancer therapies and immunotherapies that show efficacy in preclinical animal models ultimately receive approval following phase III clinical trials. This high attrition rate is mainly attributed to the limited translational relevance of existing models. Animal systems often fail to replicate the heterogeneity and complexity of human tumors, particularly in terms of immune components, due to species-specific differences and, in many cases, compromised or simplified immune functions [11,12].

A central issue in immuno-oncology is the intricate nature of the tumor microenvironment (TME). The TME comprises not only tumor cells but also stromal cells, immune infiltrates, endothelial cells, and extracellular matrix components that collectively regulate tumor growth, immune evasion, and therapeutic response. Immunotherapies often face resistance due to the immunosuppressive features of the TME, such as the presence of regulatory T cells, myeloid-derived suppressor cells, and inhibitory cytokines [13,14]. These nuances are difficult to capture using standard preclinical models.

Even within human patients, responses to immunotherapy vary significantly [15], highlighting the need for predictive, human-relevant preclinical models. Conventional two-dimensional (2D) cell cultures fail to replicate the complex architecture and cellular interactions of the tumor microenvironment (TME) [16,17]. While animal models provide a more physiological milieu, they present ethical challenges and inherent biological differences that compromise their utility, particularly in immuno-oncology research.

To overcome these limitations, researchers have increasingly turned to more physiologically relevant in vitro systems. Initial efforts focused on the development of three-dimensional (3D) tumor spheroids, which offer improved simulation of tumor morphology and biochemical gradients compared to 2D cultures [16]. These models offer enhanced insights into tumor proliferation, drug penetration, and oxygenation gradients. However, they still lack several key features necessary for studying immune-oncology applications, such as dynamic perfusion, immune–stromal–tumor interactions, and tissue-level architecture. To partially address these limitations, researchers have turned to multicellular spheroid models, which offer improved representation of the tumor microenvironment (TME) compared to traditional two-dimensional cultures. By incorporating multiple cell types, multicellular spheroid models better represent the TME and extracellular matrix (ECM), allowing for more realistic investigations of immune cell infiltration and tumor invasion [18]. These models also offer a superior platform for evaluating drug delivery and efficacy compared to 2D cultures, where complex cell–cell and cell–matrix interactions are absent [18,19]. For instance, a study using both epithelial and menenchymal-like (EMT) A549 lung adenocarcinoma cells demonstrated the formation of N-cadherin/E-cadherin adhesion complexes, highlighting the role of ECM crosstalk. This interplay with the ECM is critical for designing biomimetic models that capture tumor heterogeneity [19]. Furthermore, communication within tumor complexes often occurs via the release of soluble factors. In co-culture systems involving breast cancer cells and fibroblasts, for example, cells released vimentin, matrix metalloproteinases (MMPs), and platelet-derived growth factor (PDGF), which are factors known to influence tumor progression and the surrounding stroma [18].

Microphysiological systems (MPSs) refer to in vitro models designed to replicate complex human physiologies on a small scale. This term applies to organ-on-a-chip (OoC) technologies, as well as organoids and spheroids [20]. These platforms aim to recreate the physical, chemical, and biological conditions of human tissues, allowing the integration of vascular networks, immune cells, and stromal components, and are emerging as next-generation tools for cancer research. OoC systems allow researchers to control factors like gradient formation and cellular communication within a perfused, microfluidic environment that closely mimics in vivo physiology [21,22].

Recent studies have demonstrated the capacity of OoC models to incorporate T cells, CAR-T therapies, and cytokine signaling components, making them particularly well-suited for studying immunotherapeutic responses. OoC platforms have been used to model T cell extravasation across endothelium, CAR-T cytotoxicity against 3D tumor spheroids, and immune checkpoint blockade within vascularized microenvironments [23,24]. These systems support real-time imaging, scalability, and personalized testing while reducing reliance on animal experimentation, which is an increasingly important consideration aligned with evolving regulatory and ethical standards.

This review aims to highlight the transformative role of organ-on-a-chip (OoC) technologies in the field of cancer immunotherapy. It evaluates recent advances, examines their applications across various cancer types, and outlines how these platforms are being utilized to model immune–tumor interactions, evaluate immunotherapeutic efficacy, and improve preclinical drug testing strategies. Seven representative cancer models were examined to illustrate the most common approaches replicating complex tumor architecture and immune interactions in organ-on-a-chip and 3D culture systems. The models were selected for their methodological diversity and recent publication date, primarily within the last five years, providing complementary perspectives on cytokine signaling, vascular infiltration, and 3D tumor–immune co-culture. A comprehensive point-by-point description of the models is provided as Supplementary Information (Table S1: Organ-on-a-chip_Cancer_Immune_Research). Each model was selected for the distinct physiological features it replicates, such as immune infiltration, cytokine gradients, vascular perfusion, and multicellular organization, highlighting their relevance to preclinical immunotherapy research.

2. Organ-on-a-Chip Models for Cancer Immunotherapy

2.1. Technological Overview

OoC platforms are engineered microfluidic systems designed to replicate key aspects of human tissue architecture and function within a controlled, perfusable microenvironment. These systems enable the co-culture of different cell types, such as tumor and immune cells, under dynamic flow conditions, allowing researchers to study cellular interactions, cytokine signaling, and the effects of applied therapies in a more physiologically relevant context [24,25]. All models examined in this review simulate immune–tumor interactions and include features such as fluid flow, compartmentalization, and immune cell tracking. OoC platforms can be tailored to specific research needs, depending on the organ and cancer type being studied [25].

Most systems use polydimethylsiloxane (PDMS)-based microfluidic chips, mainly because PDMS is biocompatible, optically clear, and easy to work with [26,27]. In some platforms, polyethylene glycol diacrylate (PEGDA) hydrogels are used to provide a matrix in which cells and spheroids can grow in a 3D structure. These hydrogels have adjustable mechanical properties, helping to mimic the physical characteristics of the tumor environment [28]. Other approaches combine hybrid spheroids, consisting of multiple cell types to better replicate the tumor microenvironment (TME), with microfluidic chips to study spatially organized tumor–immune interactions [28,29,30].

One major advantage of these platforms is their ability to incorporate different immune cell types into co-culture with tumor cells. This allows researchers to study immune cell infiltration, cytotoxicity, and cytokine-influenced immune behavior of CD4+ T cells, macrophages, monocytes, and even CAR-T cells.

These models utilize perfusion to apply physiological shear stress, support nutrient exchange, and mimic immune cell migration (Figure 1). Perfusion durations range from 48 h up to eight days [22,28,29,30,31,32,33,34,35,36,37,38]. In systems not focused on drug testing, continuous perfusion has been maintained for up to one month, enabling both short-term interaction studies and long-term observation of immune–tumor dynamics [39]. Perfusion is often achieved through gravity-driven flow, using a rocker platform that tilts the chip and causes fluid to move between reservoirs. This setup offers a relatively simple and low-maintenance solution, avoiding the need for sterile connections to external pumps. While it does not fully replicate vascular flow, it still generates shear stress and supports transport processes [35,40,41]. Some systems do not use active flow at all, instead relying on passive diffusion through Matrigel membranes or across microchannels. In these cases, immune cell migration and media transport appear to be driven by chemotaxis [33,34,37]. Other platforms use syringe pumps to create unidirectional flow through continuous withdrawal at a defined rate [28,29,38,42,43]. More advanced systems incorporate closed-loop circulation with peristaltic pumps to better simulate dynamic physiological conditions [22,36,39].

Beyond flow control, channel architecture is a critical factor in determining physiological relevance. Common formats include dual-channel systems, which are separated by porous membranes that mimic vascular–tissue interfaces [31,33,37,40]; single-channel designs, which are used for tumor spheroids [30]; and multichamber systems, which allow for sequential tissue zones [34,37,42]. These configurations enable the modeling of tissue barriers, immune extravasation, and inter-tissue signaling.

Readout techniques used across these models include live-cell imaging with fluorescence and confocal microscopy to track immune and tumor cell behavior [29,30,33,35,37,40,42,43,44], as well as immunofluorescence for endpoint characterization [19,22,30,31,36,44]. Cytokine release is often quantified using ELISA or multiplex bead assays (e.g., IFN-γ, IL-6, and TNF-α), and flow cytometry (FACS) is used to evaluate immune cell activation and exhaustion markers (e.g., PD-1 and CTLA-4) as well as changes in immune subtypes [22,29,30,31,36,37,38,40,43,44]. Some platforms enable real-time monitoring of immune cell infiltration into 3D tumor spheroids, cytotoxic activity, and the dynamics of checkpoint inhibitor responses. Together, these tools offer multi-parametric insight into immune–tumor interactions under dynamic, physiologically relevant conditions.

2.2. Current Industrial Developments and Companies Commercializing MPSs in 2025

Considering their novelty, commercial OoC applications of immune-competent models are only now emerging, as reflected in a growing number of industrial platforms designed to replicate immune–tumor interactions (

Table 1. Immune-competent models based on commercial OoC platforms.

Company	Platform	Model Description	Ref.
TissUse	HUMIMIC Chip2	Human lymph node-on-chip model with integrated lymphatic vasculature and recirculating human T- and dendritic cells, enabling the study of lymphatic–endothelial interactions, immune cell trafficking, and antigen-specific immune responses.	[45]
Mimetas	Organoplate	Platform using 64 perfused HUVEC tubules to model endothelial inflammation under exposure to cytokines and immune cells, capturing barrier disruption and morphological change for studying vascular inflammation.	[46]
Hesperos	3-Organ system	Human multi-organ immune system-on-a-chip with recirculating monocytic cells simulating both targeted and systemic immune responses, enabling assessment of tissue-specific infiltration, cytokine profiles, and immune-mediated toxicity.	[47]
Cherry Biotech	Cubix	Multi-well microfluidic adaptor integrated with reconstructed human skin and flowing monocyte-like immune cells, enabling dynamic, skin-on-chip culture with control over gas and flow conditions, supporting immune activation studies.	[48]
Alveolix	AXLung-on-Chip System	Lung-on-chip model combining human pulmonary endothelial cells, circulating peripheral blood mononuclear cells, and mechanical breathing motion reveals that immune–endothelial interactions under lipopolysaccharide challenge significantly exacerbate inflammation and barrier disruption.	[49]
AIM Biotech	idenTx	RCC-on-a-chip platform recreating tumor spheroids in a collagen extracellular matrix, enabling assessment of drug responses and tumor cell migration, incorporating engineered human cytotoxic T lymphocytes to study antigen-specific immune-mediated tumor killing.	[30]

).

2.3. Application by Tumor Type

2.3.1. Lung Cancer

With around 2.2 million newly diagnosed cases and 1.8 million deaths in 2020, lung cancer ranks as the second most commonly diagnosed cancer worldwide and remains the leading cause of cancer-related death. More deaths occurred due to lung cancer than from breast, colorectal, and prostate cancers combined. The disease disproportionately affects individuals from lower socioeconomic backgrounds. Although about 80% of all lung cancers are linked to tobacco use, the disease also ranks among the top ten cancer-related causes of death in non-smokers globally [1,35,50].

The poor prognosis is largely due to late-stage diagnosis, as most cases are identified when treatment options are already limited. Over the past two decades, however, advances in tumor biology have led to the development of targeted therapies and immunotherapeutic approaches, including immune checkpoint inhibitors such as nivolumab and pembrolizumab. These therapies have significantly improved survival, especially in patients with advanced non-small cell lung carcinoma (NSCLC), and are now firmly established in clinical practice. Improvements in screening, particularly through low-dose computed tomography (LDCT), have also contributed to earlier diagnosis in high-risk populations [1,35,50].

Conventional 2D models fail to replicate the multicellular epithelium required to mimic the pulmonic architecture, and several microfluidic organ-on-a-chip (OoC) models have been developed to recreate aspects of the lung tumor microenvironment and provide more physiologically relevant tools for studying immunotherapeutic interventions [51].

One such model, described by Sardarabadi et al. [42], used a PDMS-based chip bonded to glass, with a gelatin methacryloyl (GelMA) hydrogel to embed cells in a three-dimensional matrix. The authors co-cultured A549 lung carcinoma cells with Jurkat T cells to simulate the tumor–immune microenvironment. A central IL-6 reservoir, flanked by parallel flow channels, was used to establish a chemotactic gradient. This setup allowed the researchers to observe real-time migration of T cells under flow conditions and analyze cytokine responses using ELISA. Interestingly, IL-6 was shown to have a dual role: it initially promoted T-cell activation and migration, but subsequently led to an upregulation of PD-L1 expression on tumor cells. While no drugs were tested in this particular study, the model provides a valuable foundation for future evaluations of cytokine-modulating agents and checkpoint inhibitors. The ability to monitor immune activation and exhaustion through cytokine profiling could be especially useful in preclinical screening.

Another MPS focused on NSCLC (HCC0827) employed a 3-lane chip developed by MIMETAS, incorporating a perfusable 3D endothelium. Endothelial cells (HUVECs) were cultured next to tumor cells, separated by a collagen barrier, to recreate the vascular interface. T-cells were introduced into the endothelial compartment and allowed to migrate toward the tumor under continuous perfusion. The platform enabled detailed observation of T-cell infiltration, a process strongly associated with improved clinical outcomes in immunotherapy. Although the system does not fully capture the complexity of lung tissue, it offers a reproducible and scalable method for assessing therapies that aim to enhance immune access to solid tumors [35].

Both OoC platforms demonstrate distinct strategies for modeling lung cancer immune dynamics in a setting that allows for precise manipulation of variables such as flow, cytokine gradients, and tissue architecture. One emphasized cytokine-driven migration within the tumor matrix, while the other focused on trans-endothelial infiltration. Each approach highlights specific mechanistic insights relevant for immunotherapy development in lung cancer.

2.3.2. Breast Cancer

Breast cancer is the most commonly diagnosed carcinoma among women worldwide. However, it also affects men, although at a significantly lower rate. It remains a significant global health concern. Statistically, 1 in 20 women develop breast cancer during their lifespan. Especially in countries with a low human development index, it is predicted that by 2050, 3.2 million new cases and 1.1 million deaths will occur due to this disease [52].

Despite all the developments in early diagnoses and therapy, the development of effective and safe drugs is a crucial challenge. Breast cancer is a heterogeneous disease with various molecular subtypes, including ER-positive/HER2-negative and triple-negative breast cancer (TNBC). Every subtype has its own biological properties that require different therapeutic approaches [53,54]. Triple-negative breast cancer is very aggressive and hard to treat, whereas ER-positive tumors react well to hormonal therapies. The therapy for breast cancer underwent significant development during recent years. New approaches target the hormone sensitivity of breast cancer. Treatments like CDK4&-inhibitors, SERD-Therapies (which degrade the selective estrogen receptor), and immunotherapy improved survival rates and provided longer life spans while simultaneously reducing side effects, especially for metastasized breast cancer [53,55,56].

Organ-on-chip technologies have also made it possible to replicate the complexity of breast cancer in vitro, capturing key features of the tumor microenvironment and disease progression. A remarkable example developed by Lee et al. is a dual-chip model that integrates breast carcinoma and cardiac tissue to study the cardiotoxic side effects of chemotherapy [44].

To support personalized CAR-T cell testing, Maulana et al. [43] developed a PDMS-based microfluidic platform featuring perfusable endothelialized channels next to chambers with tumor cells. These chambers contained either TNBC spheroids (MDA-MB-231) or patient-derived organoids (PDOs) embedded in a dextran-based hydrogel. CAR-T cells, engineered to recognize the Receptor Tyrosine Kinase-Like Orphan Receptor 1 (ROR1), which is overexpressed in TNBC, were continuously perfused through the endothelial channel, mimicking extravasation into the tumor site. Functional responses were evaluated using live imaging, immunofluorescence, flow cytometry, and cytokine quantification (e.g., IFN-γ, IL-2, granzyme B). The system demonstrated target-specific T-cell infiltration and tumor lysis while also capturing cytokine release syndrome (CRS)-like responses. Importantly, the use of dasatinib effectively reduced cytokine levels without compromising CAR-T cell viability. This study highlights the potential of microfluidic OoC models for evaluating both the efficacy and safety of patient-specific immunotherapies.

In another study, Berger Fridman et al. [31] established a high-throughput microfluidic platform using alginate-based hydrogels to recreate inflammatory and immunosuppressive tumor environments. By co-culturing breast tumor cells, fibroblasts, T cells, and macrophages, the platform was able to simulate dynamic shifts within the TME. Immunosuppressive shifting of macrophages to tumor-associated macrophages (TAMs) and changes in cytokine secretion were observed depending on platform configuration. Proteomic analysis revealed expression patterns that closely resemble those found in vivo, supporting the model’s relevance for studying immune cell regulation and TME progression.

Lee et al. [44] approached breast cancer modeling through a dual-organ microfluidic platform integrating breast cancer spheroids (SK-BR-3) with induced pluripotent stem cell-derived cardiac spheroids. The system allowed real-time monitoring of HER2+ tumors and cardiac responses to the drug doxorubicin using electrochemical aptasensors. While not focused on immunotherapy, the model demonstrated the different drug sensitivity in cancer versus healthy and fibrotic cardiac tissue, highlighting its value in predicting off-target cardiotoxicity of anticancer treatments in vulnerable patient populations.

A vascularized breast cancer-on-chip model developed by Boussommier-Calleja et al. [37] further investigated the relevance of immune integration by introducing primary human monocytes into a 3D microfluidic system. Tumor cells (MDA-MB-231) were seeded alongside HUVECs and fibroblasts in a fibrin-based matrix, forming perfusable microvessels. Monocytes were shown to significantly reduce tumor cell extravasation, not through direct contact, but likely via paracrine signaling involving factors such as IL-8 and CCL2. These findings emphasize the value of immune-integrated microfluidic models for investigating metastatic processes and tumor–immune interactions.

Together, these models reflect the diversity and adaptability of OoC technologies in breast cancer research. From simulating tumor–immune interactions and T-cell infiltration to evaluating patient-specific CAR-T responses or off-target toxicities, these platforms enable mechanistic insight and therapeutic evaluation under human-relevant conditions.

2.3.3. Melanoma

Melanoma is the most aggressive form of skin cancer and represents a significant health burden. Although it accounts for only 1% of all skin cancers, it causes 75% of skin cancer deaths [57,58].

The global incidence of melanoma has risen sharply in recent decades, especially in countries with populations with predominantly lighter skin tones. In 2020, 325,000 new cases were diagnosed worldwide, an increase of 41% compared to 2012. The highest rates will be seen in Australia and New Zealand, followed by North America and Europe [57].

The prognosis is highly dependent on the stage of the disease. Localized melanoma, which accounts for 78% of cases, has a 5-year survival rate of 99.6%, while the survival rate for metastatic stage IV melanoma is around 35%, despite recent advances. These improvements are due to the introduction of targeted therapies such as BRAF and MEK inhibitors and immunotherapies with checkpoint inhibitors such as anti-PD-1 and PD-L1. In addition, new approaches such as CAR T-cell therapy are being tested. Despite these advances, challenges remain, particularly in dealing with resistance and metastatic disease, such as brain metastases [58].

Preclinical models, such as animal models, are often insufficient to reproduce the complex physiology of melanoma. Consequently, emerging tools like organ-on-chip systems and 3D cultures are essential for building more accurate and translational melanoma models.

In a study by Agliari et al. [33], a microfluidic co-culture model was introduced to study the migration behavior of immune cells in response to melanoma cells. The system consisted of two chambers, one containing immune cells and the other seeded with melanoma cells. Both were connected by microchannels to mimic tissue architecture and allow real-time observation of immune cell migration. The study showed that wild-type (WT) immune cells migrated towards melanoma cells by testing spleen-derived immune cells from both wild-type and IRF-8 knockout (KO) mice. WT cells showed collective and chemotactic movement, whereas KO cells moved randomly and failed to infiltrate the tumor compartment. The work demonstrated the critical role of IRF-8 in coordinating immune cell behavior and showed how immune functionality can be measured through movement analysis (e.g., migration path, movement behavior, and directionality). The system allows quantitative analysis of immune behavior and provides a useful platform for testing therapies designed to enhance immune infiltration.

Complementing this 2D dynamic system, another study by Seo et al. [29] developed a 3D mixed spheroid model in which tumor cells were co-cultured with immune cells in a scaffold-free environment to more closely mimic in vivo tumor composition. This approach allowed the study of immune infiltration, cytotoxicity, and cytokine-driven transformation of tumor spheroids. Real-time imaging showed how immune pressure not only suppressed tumor cell viability but also altered the structure of the spheroids, reflecting the interplay between immune activation and tumor adaptation. The hybrid model is adaptable in terms of matrix composition and cell types and is suitable for testing a variety of immunomodulatory therapies, including checkpoint inhibitors and CAR-T cells.

Together, these two systems illustrate complementary strategies for modelling melanoma–immune interactions. The microfluidic chip allows real-time tracking and quantitative analysis of immune migration, while the spheroid-based system illustrates the architectural complexity of tumor–immune dynamics. Both highlight the value of immune-competent in vitro models for studying mechanisms of immune evasion and evaluating the efficacy of immunotherapeutic agents in melanoma.

2.3.4. Renal Cell Carcinoma

Renal cell carcinoma (RCC) accounts for approximately 3% of all malignant tumors and, although relatively rare, represents a clinically significant entity in oncology. In Germany alone, approximately 8350 new cases in men and 4059 in women were reported in 2019, corresponding to age-standardized incidence rates of 17.7 and 7.3 per 100,000, respectively [59]. Although incidence rates have remained relatively stable since the 1990s, mortality has decreased slightly due to advances in early detection and treatment [60].

RCC is histologically and molecularly heterogeneous, comprising several distinct subtypes with different prognoses and responses to treatment. The most common form is clear cell RCC (ccRCC), which accounts for 55–60% of cases and is often associated with VHL gene mutations and activation of hypoxia-related signaling pathways [61].

Papillary RCC (pRCC), which accounts for 9–17% of cases, is often associated with MET gene alterations, while the chromophobe subtype (chRCC), although rare (~6%), is characterized by mitochondrial-rich tumor cells. Other variants, such as sarcomatoid differentiation and collecting duct carcinoma, are associated with a particularly poor prognosis [60].

Survival outcomes vary widely depending on the stage of the disease. While localized RCC (UICC stage I) is associated with a five-year survival rate of over 95%, the prognosis drops dramatically to around 17% in cases of distant metastasis (UICC IV). Approximately 40% of RCCs are detected at an early stage, but approximately 10% are already metastatic at the time of diagnosis. The overall relative five-year survival rate is 77% for men and 79% for women [60].

Despite advances in targeted therapies—including tyrosine kinase inhibitors—and immunotherapy with checkpoint inhibitors, the treatment of metastatic RCC remains a clinical challenge. Intratumoral heterogeneity, activation of resistance pathways such as mTOR signaling, and a lack of predictive biomarkers significantly limit long-term therapeutic success [61].

Several organ-on-chip models have been developed that replicate key aspects of the RCC phenotype and are used to study its pathophysiology and drug resistance (Figure 2).

A study by Miller et al. [30] introduced a 3D RCC spheroid model cultured in a microphysiological chip. Using A498 and RCC4 cell lines embedded in a collagen matrix, the system mimicked essential features of the tumor microenvironment, including hypoxia, collective migration, and gene expression patterns consistent with in vivo RCC tissue. The platform enabled detailed drug response analysis. For example, bortezomib treatment resulted in significantly higher cytotoxicity in 3D cultures compared to 2D models, underscoring the physiological relevance of the chip. In addition, the model revealed migratory tumor behavior associated with EMT-like gene expression, enabling screening of anti-metastatic compounds such as latrunculin A and AT13148. The system also supported immune-related applications, demonstrating that ROR1-targeted CAR T cells could selectively kill RCC spheroids, although extracellular matrix penetration was a barrier. This highlights the potential of this model for testing both cytotoxic and immunomodulatory therapies in RCC.

A complementary approach was developed by Somova et al. [22,62], who designed a multi-compartment microphysiological system that co-cultured RCC cells (Caki-1) with healthy renal epithelium (RPTEC/TERT1). In their approach, tumor and healthy tissues were cultured in spatially separated but perfused chambers. Exposure to tumor-conditioned medium induced cytokine expression (IL-8, TNF-α) and metabolic reprogramming, reflecting tumor–host interactions. While immune cells were not added, the chip is compatible with future immune co-culture configurations, offering a basis for evaluating tumor–host and immunotherapy–host interactions.

Taken together, these studies highlight the growing value of MPSs in RCC research. They offer a reliable way to model the tumor microenvironment, enabling long-term perfused co-cultures of tumor and surrounding tissue. These platforms also support real-time analysis of cytotoxicity, metabolism, and cytokine signaling, making it possible to assess how drugs affect both the tumor and its microenvironment.

There are still limitations, especially when it comes to capturing the full complexity of the immune system or simulating systemic immune responses and vascular trafficking [22,30,62]. Moving forward, integrating more immune components and dynamic vascular features will be essential to increase the translational potential of these models.

2.3.5. Diffuse Large B-Cell Lymphoma

DLBCL is the most common form of non-Hodgkin lymphoma, accounting for 30–40% of all cases, and represents an aggressive, life-threatening disease. The annual incidence is 7 per 100,000 people, with a mean age of onset of 64 years and a higher prevalence in men. Without treatment, the disease progresses rapidly, emphasizing the urgent need for effective therapeutic strategies [63].

DLBCL is a highly heterogeneous disease with multiple molecular subtypes. The most clinically relevant are the germinal center B-cell-like (GCB) and activated B-cell-like (ABC) subtypes, which differ in both genetic signature and clinical behavior. Genetic alterations such as translocations involving BCL2, BCL6, or MYC play central roles in disease pathogenesis and influence treatment response. The current standard of care is the R-CHOP regimen (rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone), which is potentially curative for many patients. However, relapse occurs in approximately 40% of cases, and the prognosis remains particularly poor in aggressive subtypes [64].

A study by Mannino et al. [38] introduced a lymphoma-on-a-chip platform designed to replicate key features of the DLBCL tumor microenvironment (Figure 3). The model uses a hyaluronic acid-based hydrogel embedded with tumor B-cells, T-cells, and macrophages, along with a perfusable, endothelialized microchannel to simulate tumor vasculature. This configuration enables the spatiotemporal analysis of drug delivery under dynamic flow conditions. The chip supports fluorescence-based diffusion tracking, flow cytometry analysis, and incorporates spatially distinct regions that mimic both tumor tissue and healthy lymphatic environments.

As a proof of concept for therapeutic testing, the system was treated with an anti-CSF-1R antibody, resulting in selective macrophage depletion within the tumor compartment. This demonstrated the model’s capability to evaluate drug efficacy in real time and under immune-competent, physiologically relevant conditions. The inclusion of physically separated compartments also allowed for the analysis of localized versus systemic immune effects, which is a feature not commonly addressed in traditional models.

While the system offers advanced spatial and functional complexity, several technical limitations were noted. For example, the removal of the steel wire used in fabrication occasionally disrupted the hydrogel structure, leading to variability in channel geometry. Additionally, despite its sophistication, the model does not yet replicate systemic immune circulation or multiorgan interactions [38].

A second study by Foxall et al. [32] focused on developing a scaffold-free 3D tumor microenvironment (TME) model for B-cell lymphoma to assess antibody-based therapies. This model was particularly suited to evaluating monoclonal antibody responses, including responses to rituximab. Compared to standard 2D cultures, it more accurately reproduced tumor morphology and endothelial barrier integrity, revealing resistance mechanisms that align with in vivo behavior. However, the absence of vasculature and dynamic perfusion limited its ability to simulate drug delivery kinetics and immune cell migration.

Together, these two approaches reflect the increasing importance of 3D and microfluidic models in lymphoma research. While one emphasizes spatial control and vascularized architecture, the other demonstrates the value of preserving tumor structure for therapeutic assessment. Both offer greater precision in modeling tumor–immune interactions than conventional systems and underscore the need for continued development toward more physiologically complete platforms.

2.3.6. Hepatocellular Carcinoma

Hepatocellular carcinoma (HCC) is the most common primary liver cancer, accounting for more than 90% of cases, with a global incidence of 500,000 to 1 million new diagnoses per year [65,66]. Its pathogenesis is closely associated with chronic liver diseases such as viral hepatitis (HBV/HCV), cirrhosis, and metabolic disorders. The tumor microenvironment is characterized by hypoxia, angiogenic signaling (e.g., VEGF, HIF-1α), and immune cell infiltration [67,68]. These complex interactions make HCC a particularly challenging target for conventional therapies, especially given that current in vivo models poorly recapitulate human liver physiology and immune context.

Treatment strategies for HCC are highly stage dependent. Early-stage disease is typically treated by resection, liver transplantation, or radiofrequency ablation for tumors less than 3 cm [69,70]. Intermediate stages are treated with transarterial chemoembolization (TACE), with a median survival of 20–26 months [69]. Emerging strategies now combine TACE with immunotherapy to improve efficacy [71]. For advanced HCC, first-line treatments include atezolizumab/bevacizumab (anti-PD-L1 + anti-VEGF) or durvalumab/tremelimumab (anti-PD-L1 + anti-CTLA-4) combinations [71]. Targeted therapies such as lenvatinib and sorafenib continue to be used, although resistance, often due to reactivation of the MAPK pathway, continues to limit long-term success [72].

In order to enhance understanding of immune cell trafficking in the HCC tumor microenvironment, Kennedy et al. [34] developed a two-channel liver-on-a-chip model that reproduces key features of hepatic sinusoidal flow and endothelial cell interactions. The study’s primary focus is on the adhesion and migration of immune cells across primary human liver sinusoidal endothelial cells (LSEC) under physiological shear stress, with implications for the testing of immunotherapies and the delivery of chimeric antigen receptor (CAR)-T cells. The model is constructed using the Emulate Liver-Chip platform, a microfluidic device with two parallel channels separated by a porous membrane. LSECs are cultured in the lower (endothelial) channel, while Huh-7 hepatoma cells are cultured in the upper (epithelial) channel. Perfusion is applied to mimic the sinusoidal shear stress of the liver (0.05 Pa) [34]. Following a 24 h period of cytokine stimulation with TNF-α and IFN-γ, the LSEC layer showed increased immune activation and morphological changes. Pre-labeled peripheral blood mononuclear cells (PBMCs) were introduced into the chip and exposed to the endothelial layer. Confocal and multiphoton imaging techniques were used to visualize immune cell adhesion and migration across the LSEC. It was observed that cytokine treatment significantly enhanced immune cell adhesion, as compared to the control group [34].

This system enables the real-time study of leukocyte recruitment and migration across the human liver endothelium. This system has the potential to test the efficacy of immunomodulatory therapies, which include checkpoint inhibitors and cell therapies such as CAR-T or CAR macrophages, in enhancing immune infiltration into HCC tissues [34].

2.3.7. Prostate Cancer

Prostate cancer (PCa) is the most common cancer in men in the USA and Europe, with a projected incidence of over 313,000 new cases and 35,000 deaths in the US for the year 2025 [73].

Therapeutic strategies for prostate carcinoma vary depending on the stage of the disease. For localized, low-risk tumors, active surveillance is often the recommended approach. In intermediate- and high-risk patients, radical prostatectomy is considered standard, often complemented by radiation therapy [74].

For advanced and metastatic prostate cancer, androgen deprivation therapy (ADT) remains the foundation of treatment. In many cases, ADT is initiated in combination with newer hormonal agents such as abiraterone or enzalutamide to improve outcomes. Chemotherapy may also be considered [75].

In castration-resistant prostate cancer (CRPC), where the disease progresses despite low testosterone levels, targeted therapies have emerged. Olaparib is used in patients with BRCA1/2 mutations, while radioligand therapy with ¹⁷⁷Lu-PSMA-617 offers a treatment option for patients with PSMA-positive metastases [76,77].

PCa research is hampered by the limited availability of robust and representative cell lines as well as highly inefficient preclinical animal models [78].

To address these limitations, Peng et al. [28] developed a sophisticated microphysiological tumor model using PEGDA hydrogel microbeads to recreate a tunable, 3D tumor microenvironment containing prostate cancer cells and fibroblast-associated stromal components. The authors co-cultured PSCA-expressing PC3 prostate cancer cells with HT1080 fibrosarcoma cells modified to express fibroblast activation protein (FAP), a stromal marker associated with immune suppression. These composite spheroids were encapsulated in PEGDA beads with mechanical properties that mimic the tissue stiffness of common metastatic sites such as the liver and lung.

This model enabled long-term spheroid viability, and was used to evaluate UniCAR T cell therapy targeting both PSCA (tumor antigen) and FAP (stromal antigen). The study showed that dual-targeting UniCAR T cells could efficiently infiltrate the matrix, accumulate on the spheroid surface, and induce potent tumor cell killing. Cytokine analysis (e.g., IFN-γ, TNF-α) and granzyme B immunostaining confirmed T cell activation and cytotoxicity, while migration assays tracked T cell behavior in real time [28].

A study by Xie et al. [40] used a pump-free, dual-channel microfluidic chip to study interactions between natural killer (NK) cells and DU145 prostate tumor spheroids. The chip’s gravity-driven design created unidirectional fluid flow to mimic physiological perfusion without external pumps. Tumor spheroids sat in Matrigel-coated paper stacks within the chip, and NK-92 cells were introduced into adjacent channels. The flow conditions allowed NK-92 cells to migrate and infiltrate the spheroids more, and the flow influenced the cells’ localization and activation.

Static and perfused conditions revealed that dynamic culture better reflected in vivo NK cell phenotypes, particularly the transition to tumor-infiltrating NK cell (TINK) subtypes, as shown by changes in CD56/CD16 marker expression. The setup also allowed separation and phenotypic analysis of free versus infiltrated NK cells, highlighting the chip’s utility for prostate tumor immune dynamics [40].

In a complementary approach, Padmyastuti et al. [36] performed a study using the HUMIMIC Chip platform (TissUse) to investigate molecular and secretory changes in prostate cancer models under 3D and microfluidic conditions (Figure 3). LNCaP and PC3 prostate cancer cells were cultured in both 3D static and within the perfused HUMIMIC platforms. The model sustained PSA secretion in LNCaP cells for over 10 days and induced changes in microRNA expression profiles compared to 2D cultures. MicroRNAs linked to castration-resistant prostate cancer were upregulated under MPS conditions, suggesting that dynamic flow impacts molecular traits relevant to disease progression. PSA and PSMA levels were more robustly expressed in microfluidic culture, indicating that the system supports physiological cell behavior.

Together, these platforms demonstrate the diverse capabilities of prostate cancer–focused OoC systems. The PEGDA-based model emphasizes dual-targeted CAR-T cell immunotherapy, the pump-free chip reveals NK cell behavior under dynamic conditions, and the dynamic MPSs showcase phenotypic changes and biomarker expression. Each contributes valuable insights toward building predictive, immune-integrated models for prostate cancer drug development and immunotherapy testing.

3. Key Experimental Approaches Across Models

Across the different cancer-specific OoC and microphysiological systems reviewed, several recurring experimental strategies and design principles can be observed. These approaches define the flexibility and translational potential of in vitro immuno-oncology modeling.

3.1. Cell Sources and Cell Types

The majority of models include a combination of immortalized tumor cell lines (e.g., A549 for NSCLC, Huh-7 for HCC, CAKI-1 for renal cell carcinoma), endothelial cells (often HUVECs), and stromal cells such as fibroblasts. The integration of immune cells is critical to reflect the specific mechanisms of immunotherapy. These include CD4+ T-cells, Jurkat cells, CAR-T cells, monocytes, macrophages, and peripheral blood mononuclear cells (PBMCs). In certain models, tumor spheroids are constructed using PEGDA hydrogels or Matrigel to mimic the 3D tumor architecture [28].

3.2. Immune Cell Integration Methods

The integration of immune cells can be achieved by direct embedding in hydrogels, perfusion through adjacent channels, or co-culture in separate but interconnected compartments (Figure 4). For instance, the addition of CAR-T cells is frequently achieved through a recirculating perfusion loop or a syringe–pump system to simulate vascular infiltration [34,43]. Monocytes and neutrophils are perfused through channels to study transmigration or immune suppression [31,35,37].

3.3. Model Viability and Stability

A variety of techniques are used for model characterization and response assessment. For imaging, confocal microscopy [33,37,39,41,42,79], live-cell fluorescence imaging, and immunofluorescence are employed to track cell localization, toxicity, and infiltration [29,31,32,38,41,42,80]. For cytokine analysis, ELISA and multiplex assays are frequently utilized, for instance, for IFN-γ, TNF-α, and IL-6. FACS and immunostaining are employed to evaluate activation markers or PD-1/PD-L1 expression. Spheroids are analyzed for viability, area, and morphological changes in 3D cultures [29,31,42,43].

3.4. Engineering Innovations

Engineering innovations have led to the development of novel platforms that enhance physiological relevance in vitro. These include PEGDA hydrogels with tunable stiffness to mimic matrix mechanics and cytokine gradient systems to simulate inflammation-driven chemotaxis [28]. The drive to better mimic the tumor microenvironment (TME) was evident in nearly all cancer types reviewed. The incorporation of diverse cell types into spheroids appears to be the next evolution in simplified 3D culture systems [28,29,31,34,42]. Dual-compartment chips offer a practical means of replicating different anatomical or physiological compartments, allowing the study of local versus systemic immune responses. Taken together, these experimental designs highlight the potential of organ-on-chip (OoC) platforms to model complex immune–tumor interactions under dynamic and modular conditions (

Table 2. Strengths and limitations of current OoC models and future directions.

Strengths	Current Limitations
Physiologically relevant design: Enables simulation of tissue-specific architecture, flow conditions, and mechanical stress, better reflecting the tumor microenvironment [81,82,83].	Restricted immune complexity: Most models only include one or two immune cell types, lacking full immune repertoire [82].
Immune–tumor co-culture compatibility: Supports direct interaction between tumor cells and immune components (e.g., T cells, CAR-T, macrophages) [82].	Short experimental lifespan: Co-cultures are typically viable for less than 10 days, limiting long-term studies [84].
Real-time monitoring: Optical clarity and microfluidic control allow for live-cell imaging, migration tracking, and real-time immune response evaluation [83,85].	Absence of systemic dynamics: Models lack full-body immune circulation or multiorgan crosstalk relevant in metastatic or immunomodulatory settings [82,86].
Ethically favorable: Helps reduce reliance on animal testing, aligning with regulatory efforts toward human-relevant in vitro alternatives [81,83].	Inconsistency in chip fabrication: Manual handling and variable gel properties can lead to differences in structure and flow patterns [85].
Customizable and modular: Devices can be tailored for specific cancer types, immune interactions, or treatment scenarios [82,83].	Standardization hurdles: Lack of harmonized protocols complicates cross-platform comparison and clinical translation [85,87].
Useful for immunotherapy evaluation: Some platforms allow testing of immune checkpoint inhibitors, cytokine blockers, or engineered cell therapies [82].	Immune cell adhesion and blockage: Immune cells can adhere to microchannel surfaces or become blocked in MPSs [86,87].

).

OoC have introduced new levels of physiological relevance and experimental control, overcoming current limitations will be key to advancing their translational value. A major challenge remains the limited immune complexity, as most models incorporate only one or two immune cell types [88]. A potential solution lies in the integration of some form of bone marrow-on-a-chip components or immune organoids, which could provide a more diverse and functional immune repertoire within the system [39,89]. Expanding this to include a broader immune repertoire, such as a combination of dendritic cells, NK cells, and B cells, could be essential for more detailed studies of immune suppression, activation thresholds, interactions, and immune escape mechanisms. In addition, immune memory, exhaustion, and chronic activation states remain difficult to model in short-term cultures.

Similarly, the limited biological complexity of tumor models can be partially overcome by incorporating patient-derived organoids that contain stromal, vascular, and autologous immune cells, allowing for more realistic tumor–immune interactions. Combined with microfluidic perfusion and real-time analytics, this could enable ex vivo immunotherapy testing customized to individual patients, supporting personalized oncology [90,91].

On the technical side, standardization of chip fabrication methods and operating protocols will be essential to improve interlaboratory reproducibility. Initiatives such as the organ-on-a-chip standards that may be developed by the European Committee for Standardization (CEN) are an important step in enabling wider adoption of these systems in both research and regulatory contexts.

The short lifespan of immune–tumor co-cultures and the lack of systemic circulation remain persistent barriers to modeling long-term immune responses and metastatic progression. These challenges could be addressed by the development of interconnected multi-organ platforms or vascularized microfluidic circuits that would allow recirculation of immune cells and study of inter-tissue immune dynamics over extended periods of time.

Current systems are optimized for acute responses, with experiments rarely extending beyond a week. However, many immunotherapeutic outcomes depend on long-term cell–cell communication, adaptive resistance, and tumor evolution. Increasing the duration and stability of co-cultures will be critical for modeling events such as checkpoint inhibitor resistance or CAR-T cell exhaustion.

Finally, some functional limitations, such as immune cell adhesion or vascular blockage, can be mitigated by better design strategies, including the use of surface coatings, optimized channel geometries, and adaptive flow control mechanisms. Addressing these technical and biological gaps will be critical to transforming OoC systems into robust tools for immunotherapy development and precision medicine [86,87].

4. Conclusions

OoC technologies have attracted considerable interest as tools for modeling cancer–immune interactions in vitro. However, many of the models described in the current literature should be considered primarily as proof-of-concept platforms. While some have tested immunotherapies or novel agents, comprehensive drug screening remains scarce. The field has clear potential for predictive drug testing, but most systems have not been developed or validated for this purpose on a large scale.

Many second-generation OoC platforms have evolved from early 3D spheroid culture approaches to incorporate controlled perfusion, spatial compartmentalization, and increasingly complex tumor microenvironments. While these systems demonstrate considerable creativity, only a subset is likely to remain relevant or scalable over the long term. Those with simplified operation, cost-effective design, and robust, reproducible outputs are most likely to transition from academic research tools to translational platforms.

While these systems do not yet replace animal models, they successfully replicate specific immune and tissue-level functions that do not require whole-organism systems. As such, they may serve as viable alternatives to xenograft models, particularly for mechanistic studies and early-stage therapeutic development. Over time and with further refinement, they could ultimately help to reduce and in some cases replace the use of animal testing in oncology altogether. The recent roll-out of the Federal Drug Administration (FDA) 3-year roadmap to reduce animal testing (April 2025) is seen as a major boost for MPS applications. It plans for a phased reduction in animal experimentation requirements in the evaluation of monoclonal antibody therapies, and subsequent expansion to biologics and small-molecule drugs (https://www.fda.gov/media/186092/download?attachment; accessed on 30 May 2025).

The growing interest in employing OoC platforms to study not only cancer immunology but also metabolic diseases, sepsis, and vaccine responses underscores their broader potential. As these tools evolve, they will become increasingly relevant for studying individualized therapies such as CAR-T cells or personalized cancer vaccines. While the field is still in its early stages, its momentum and versatility suggest that OoC models should be taken seriously as part of the future landscape of immuno-oncology research.