Comparative characteristics of in vitro models for studying angiogenesis in cardiovascular disease

Abstract

Angiogenesis is a central mechanism in the development and progression of cardiovascular diseases. Experimental approaches for studying angiogenesis vary widely, and their translational value depends strongly on model characteristics. We aimed to provide a comparative analysis of contemporary in vitro models used to study angiogenesis and to assess their potential applicability in cardiovascular medicine. Fifty-four publications by domestic and international authors were analyzed. 2D models remain accessible tools for investigating endothelial proliferation, migration, and early angiogenic responses; they are easy to implement and highly reproducible, but lack physiological relevance. 3D models better recapitulate extracellular matrix architecture and cell-cell interactions, providing higher biological fidelity at the cost of increased technical complexity and expense. Microfluidic systems reproduce hemodynamic forces and microenvironmental gradients with the highest degree of physiological relevance, but are time- and resource-intensive. Models based on induced pluripotent stem cells enable patient-specific investigations and disease modeling, although they can be limited by variability and potential instability. No single in vitro platform fully reproduces the complexity of angiogenesis. Model selection should be aligned with specific research objectives. Integrating 3D culture systems, microfluidics, and artificial intelligence-assisted analysis is particularly promising for advancing angiogenesis research in cardiovascular medicine.

Key Words: Angiogenesis; In vitro models; Cell technologies; Cell cultures; Endothelial cells

Core Tip: Experimental angiogenesis research is rapidly evolving in cardiovascular medicine. In vitro models offer precise control and stable experimental conditions with favorable cost-effectiveness compared with ex vivo and in vivo approaches. Limitations include incomplete reproduction of physiological conditions, particularly in 2D systems. Wider implementation of 3D culture, microfluidic platforms, and artificial intelligence should be considered to achieve more comprehensive and physiologically relevant studies of angiogenesis despite greater complexity and cost.

INTRODUCTION

According to the World Health Organization, cardiovascular disease remains the leading cause of death across diverse populations. Endothelial cells are central to vascular homeostasis, and numerous studies link the development and progression of cardiovascular disease to disturbances in endothelial function and angiogenesis, which makes the creation and extensive use of experimental models aimed at investigating endothelial pathology essential. A key aspect of endothelial function is the regulation of angiogenesis, defined as the formation of new blood vessels from pre-existing vascular networks, and this process is critically involved in both physiological and pathological settings[1].

Angiogenesis occurs during normal growth and tissue development, as well as under such conditions as tissue regeneration, thrombus recanalization, and the resolution of inflammatory foci. Under physiological conditions, angiogenesis proceeds in a moderate manner, whereas in tumors it is persistent and intensive, thus promoting disease progression. In myocardial and peripheral limb ischemia, therapeutic stimulation of angiogenesis is a highly desirable strategy for restoring blood flow and preserving ischemic tissues. Conversely, in atherosclerosis, intraplaque angiogenesis contributes to the instability and rupture of plaques, leading to acute thrombotic events. Furthermore, pathological angiogenesis is a major driving force of stent restenosis, since it supports neointimal tissue growth and ultimately results in vessel re-occlusion[2-4]. Given its central role in ischemia, atherosclerosis, and restenosis, there is an urgent need to elucidate the molecular mechanisms that govern angiogenesis.

Several categories of experimental approaches are used to study angiogenesis, including in vitro, ex vivo, and in vivo models[5], and progress in angiogenesis research critically depends on the availability of reliable and physiologically relevant experimental systems. Although conventional two-dimensional in vitro models are simple to implement and relatively inexpensive, they do not reproduce the three-dimensional architecture, intercellular interactions, or mechanical cues inherent to the native vascular microenvironment, which limits their predictive value. For this reason, improved in vitro models that more accurately reproduce the complexity of angiogenesis in humans are actively being developed and implemented[6-9].

This review describes and compares in vitro angiogenesis models, outlining the advantages and limitations of each type.

BASIC INFORMATION

Angiogenesis can proceed through two distinct mechanisms: Endothelial cell migration into connective tissue accompanied by basement membrane degradation, or intussusceptive capillary growth, which expands the vascular network by forming transcapillary tissue pillars[10]. Following trauma or hypoxia, vascular permeability increases due to the local release of growth factors and inflammatory mediators. This leads to vasodilation, loosening of intercellular junctions, and degradation of the basement membrane. Endothelial cells then migrate into the parenchyma through the disrupted membrane. Endothelial progenitor cells subsequently differentiate into mature endothelial cells under the influence of angiogenic factors and form immature capillary structures. Pericytes and smooth muscle cells are then recruited to stabilize these primary vascular structures, ultimately giving rise to complex three-dimensional vascular networks[11].

Endothelial cells are characterized by tight intercellular adhesion and a lack of intercellular substance, and their morphology varies substantially depending on tissue context. They differ in shape, size, orientation relative to the vessel axis, and functional properties, which reflects population heterogeneity. Experimental studies demonstrate that endothelial cells produce biologically active substances that regulate proliferation, vascular tone, coagulation, and hemostasis[12].

For many decades, endothelial cell cultures have been extensively characterized and standardized for experimental use. The availability of numerous endothelial cell lines enables researchers to select models according to specific experimental objectives based on morphological, biochemical, and culture properties. Human umbilical vein endothelial cells are most commonly used because they are inexpensive, readily isolated, and easy to maintain. In vitro models are essential for rapid acquisition of quantitative angiogenesis parameters and for evaluating factors that modulate this process.

At present, no in vitro angiogenesis model is capable of consistently reproducing all stages of vascular development. Most existing systems capture only a single stage of angiogenesis. The primary applications of in vitro models include the analysis of endothelial proliferation, migration, and differentiation[13].

2D MODELS

The most commonly used approach for studying angiogenesis in vitro is two-dimensional monolayer cell culture. In such systems, endothelial cells are grown on a flat plastic surface and receive an equal supply of nutrients and growth factors from the culture medium. Transwell-based modifications, which employ a permeable membrane to provide contact between cells on opposite sides, and Boyden chamber assays, which simulate invasion through a semipermeable membrane, are frequently used to investigate migratory and invasive properties. In addition, pseudo-two-dimensional systems examine interactions between tumor cells and normal somatic or stem cells using extracellular matrix substrates[14].

2D models are easy to implement, inexpensive, and provide highly reproducible, readily quantifiable results, making them suitable for studying basic mechanisms of vascular pathology and for primary screening of angiogenic factors[15]. Dead cells detach readily during medium changes, which simplifies analysis. However, due to their geometric and functional simplicity, 2D cultures cannot reproduce the structural complexity or physiological responses of endothelial cells observed in vivo, since they lack three-dimensional architecture and functional heterogeneity of vascular tissue[16]. Moreover, under standard 2D conditions, endothelial cell properties rapidly deteriorate, which limits their applicability for long-term studies[17].

3D MODELS

“Sandwich” models and hydrogels

The limitations of 2D cultivation can be partially overcome by adding Matrigel or another extracellular matrix component on top of the cell layer, forming “sandwich models”. Alternating layers of endothelial cells and other cell types (e.g., fibroblasts, myocytes, myoblasts, cardiomyocytes, hepatocytes, tumor cells) are interposed with cell-derived matrix or exogenous substrates such as Matrigel. Matrigel, containing laminin and type IV collagen, facilitates cell adhesion and migration. When endothelial cells are embedded in Matrigel, they display lineage-specific differentiation, moderate proliferation, and, within about 12 hours, begin to form capillary-like tubular structures in response to growth factors and matrix-bound adhesion cues[18-21].

These models allow cell anchorage in all three spatial dimensions and reproduce conditions characteristic of fibrotic foci[22]. Cells undergo stabilization and polarization, forming distinct apical and basolateral surfaces. Hydrogels can encapsulate and release bioactive substances that regulate cell differentiation. Limitations include heterogeneity in culture size and composition, dysregulated proliferation with overgrowth, accumulation of metabolic by-products, and batch-to-batch variability in hydrogel composition (matrix density and fiber architecture), which may either suppress sprouting if too dense or fail to support migration if too sparse[23-25].

Cell sheets

Cell sheet (CS) technology, developed on the basis of the sandwich cultivation approach, represents an innovative strategy in regenerative medicine aimed at recreating functional tissue structures. CS constructs are generated from mesenchymal stromal cells (MSCs) derived from adipose tissue or bone marrow together with their self-produced extracellular matrix. During cultivation, MSCs and their matrix form three-dimensional vessel-like architectures without the need for artificial scaffolds due to the intrinsic capacity of these cells for self-organization. MSCs within CS exhibit markedly enhanced pro-angiogenic and regenerative activity driven by secretion of vascular endothelial growth factor (VEGF) and fibroblast growth factor (FGF), and other vascular growth factors.

As a flexible bioengineering platform, CSs offer broad opportunities for personalized regenerative therapy and have demonstrated efficiency in promoting vascularization of multiple tissues, including cardiac tissue, where they may be used for myocardial regeneration after infarction. Unlike conventional tissue engineering approaches, CSs can generate functional capillary networks through spontaneous cellular organization, thereby improving graft survival. Future research should focus on long-term safety and efficacy evaluation as well as on the standardization of manufacturing protocols, which is critical for clinical translation. Limitations of this method include insufficient capillary maturation and structural instability associated with the absence of pericytes[25].

Spheroids

Spheroids are 3D aggregates (typically 50-500 μm) in which capillary-like networks usually appear by day 7[26]. Compared with monolayers, cells in spheroids acquire in vivo-like properties, establish intercellular contacts, exchange regulatory signals, and deposit extracellular matrix[27,28]. Hybrid spheroids composed of multiple cell types can be generated[29]. Spheroids may be formed without scaffolds using the hanging-drop technique[30] or by culturing on non-adhesive/Low-adhesive dishes[31]. These models are biologically relevant and cost-effective but require strict control of cell density and size, as diffusion limitations for oxygen and nutrients can cause central necrosis and restrict lifespan[32].

Microfluidic systems

Spheroid-based systems can be integrated with microfluidic technologies, which control fluid behavior in channels tens of micrometers wide. Microfluidic chips reduce reagent consumption, improve oxygen and nutrient delivery, extend culture lifespan, and enhance reproducibility and structural uniformity. They allow manipulation of single cells or multicellular structures under precisely controlled conditions, enabling simulation of pathological microenvironments. The main challenges are precise flow control and fabrication of microscale channels[33-35].

Organoids and organ-on-a-chip

Three-dimensional culture technologies are not limited to spheroids. Organoids represent stem cell-derived 3D structures that reproduce native tissue architecture and multilineage differentiation. Formation of organoids requires embedding cells in an extracellular matrix such as Matrigel and exposure to defined environmental and niche signals that direct self-organization[36].

Owing to their complex structure and architectural similarity to native tissue, organoids bridge the gap between in vitro systems and in vivo physiology. When derived from patient-specific cells, organoids enable the generation of individualized disease models for selecting targeted therapies. However, organoids lack an intrinsic vascular system and therefore can be generated only at micrometer-to-millimeter scale. In the absence of perfusion, nutrient diffusion from the surrounding matrix is insufficient, frequently resulting in necrosis of the inner core[37]. Additional limitations include batch-to-batch heterogeneity and incomplete maturation due to the complexity of self-differentiation, which compromises reproducibility and yields a fetal-like phenotype that limits extrapolation to adult physiology[38].

In contrast, organ-on-a-chip systems are multi-channel microfluidic 3D platforms that mimic the structural, functional, and physiological responses of entire organs. These models are widely applied in drug screening and personalized therapy development[39]. One such example is a perfused microfluidic model used to study interactions between blood components and vascular endothelium in thrombosis and atherosclerosis. The device, fabricated from polydimethylsiloxane, contains a reservoir for medium and parallel microchannels. A hydrogel precursor containing endothelial cells and fibroblasts is injected into the channels via a needle, which is subsequently removed after gelation, forming a circular lumen for cell implantation and controlled perfusion.

Heart-on-a-chip

Heart-on-a-chip (HoC) technology integrates tissue engineering and microfluidics to model cardiac function in vitro. Selecting appropriate cell types and co-culture conditions is critical. HoC platforms provide a controlled microenvironment, reproduce key myocardial functions, and can incorporate sensors for contractility and conductivity, offering a powerful platform for drug screening and cardiotoxicity prediction[40,41]. Compared with animal studies, they reduce cost and time[42], but natural extracellular matrix-like (ECM) biomaterials can be weak, degradable, and variable; their microarchitecture depends on concentration, temperature, and pH, affecting reproducibility. Synthetic matrices offer stability but reduced bioactivity[43]. Additional limitations include high cost, technical complexity, and dominant surface effects in very small liquid volumes, which may distort biochemical signals[44].

INDUCED PLURIPOTENT STEM CELL MODELS

The development of three-dimensional angiogenesis models increasingly relies on single-source systems based on human induced pluripotent stem cell (hiPSC)-derived endothelial cells, extracellular matrix scaffolds, and microfluidic devices. These patient-specific platforms enable physiologically relevant preclinical studies of endothelial-matrix interactions, vascular pathology, and drug pharmacokinetics. Three-dimensional systems more accurately reproduce cellular interactions and microenvironmental cues that regulate angiogenesis in vivo. Owing to diffusion-limited nutrient and oxygen distribution, cells within 3D constructs exhibit heterogeneous cell-cycle states; however, their native-like spatial organization preserves endothelial phenotype and function more effectively than in 2D culture.

The general approach involves encapsulating hiPSC-derived endothelial cells in a natural or synthetic biomaterial and culturing them under microfluidic conditions. The scaffold must provide an ECM-like environment that supports viability, proliferation, and function. Three-dimensional culture methods fall into two categories: Scaffold-based and scaffold-free. In scaffold-based systems, hydrogels reproduce the molecular environment of ECM in vivo and serve as a structural support that cells can remodel. Scaffold-free systems rely on endogenous ECM secretion by the cells, producing a natural environment without exogenous matrices[45].

BIOPRINTING

Bioprinting is a digitally controlled technique for fabricating 3D biological constructs by sequential deposition of biomaterials. The major modalities include extrusion-based bioprinting, droplet-based bioprinting, and laser-assisted bioprinting[46,47]. These approaches enable the fabrication of tissue-engineered constructs for studying revascularization of injured tissues or implanted scaffolds. Strategies include implantation of pre-vascularized constructs or incorporation of angiogenic factors such as VEGF, FGF, and platelet-derived growth factor to promote vessel ingrowth. Bioprinting can also be employed to create functional anastomoses or synthetic vascular grafts. However, synthetic grafts have demonstrated consistent success only for vessels ≥ 6 mm in diameter. A major limitation is that printed vessels frequently fail to fully replicate native vascular architecture, and constructs thicker than 200 μm become non-viable without prior vascularization due to diffusion constraints[48-50].

ARTIFICIAL INTELLIGENCE

Artificial intelligence (AI) provides tools for analyzing complex datasets and modeling angiogenesis. Convolutional neural networks support automated image processing and quantification in 3D models, and generative adversarial networks aid in silico design of vascular networks, accelerating development of bioprinted constructs. AI-assisted analysis enables high-throughput, objective quantification but raises questions about interpretability. In the long term, digital twins of vascular networks could simulate in vitro and in vivo responses and reduce reliance on animal testing[51-53].

DISCUSSION

2D models are the most common due to their simplicity of implementation, low cost, and ease of analysis. However, they cannot fully reflect intercellular interactions and physiological responses. Despite this, they remain indispensable for primary screening of angiogenic factors and assessment of the cytotoxicity of substances.

The development and improvement of 3D models (sandwich models, spheroids, organoids, CS) give better possibilities to study of cardiovascular pathologies, more accurately simulating pathophysiological conditions. They allow better assessment of the processes of interaction of vascular cells, vascular inflammation, and vascular aging. However, such models have a higher cost, a short service life and do not solve the problem of lack of perfusion, which limits their use in long-term experiments. 3D models effectively simulate hypoxia and revascularization, which is suitable for studying tissue ischemia and testing pharmacotherapy for cardiac protection, and also have potential for personalized medicine.

Microfluidic systems are the most promising for studying pathologies, as they can imitate hemodynamics, which brings them as close as possible to the conditions of the body. Potentially, organ-on-a-chip technologies can replace existing tests conducted on animals. The advantage of organ-on-a-chip models over in vivo testing is their lower cost and shorter time investment. One of the disadvantages of microfluidic systems is the surface effect, since the volume of liquid is very small, then the surface effects dominate over the volume effects. One of the reasons is the laminar flows that are present at the intersection of several liquids, respectively, they cannot mix sufficiently, and part of the studied liquid can be adsorbed on the internal shells, which leads to a decrease in the quality of the analysis of the effects of the substance under study[54]. Another disadvantage is the lack of portability of the systems necessary for reliable analysis of the phenomena occurring in the experiment. Also, the use of advanced models is limited by the high cost, complexity of production, and the need for specialized equipment. Microfluidic systems are most suitable for studying pathologies for the modeling of which hemodynamics is necessary: The formation of plaques in atherosclerosis, pharmacodynamics of drugs (Table 1).

Table 1 Comparative characteristics of in vitro angiogenesis models.

Ref.	View	Description	Application in cardiology and angiology	Pros of the model	Cons of the model	Speed of implementation
Galimova et al[15], 2018	2D cultivation	Cultivation occurs on a culture dish, the cells form a monolayer, all cells receive the same amount of nutrients and growth factors from the nutrient medium	Study of endothelial cell proliferation during hypoxia, screening of angiogenic factors	Low cost, ease of implementation, ease of analysis of results, high reproducibility	Lack of 3D interactions, low physiological relevance, rapid loss of functional properties of cells	< 1 week
Ballester-Beltrán et al[21], 2015	Sandwich model	Multilayer cultures with alternating cells and extracellular matrix (e.g. Matrigel)	Modeling fibrotic foci and angiogenesis under conditions of chronic inflammation	Provide cellular adhesion. Stabilizes and polarizes cells. Allows to simulate conditions occurring in fibrous lesions	Heterogeneity of structure. Variations in the composition of Matrigel from batch to batch	1-2 weeks
Laschke and Menger[27], 2017	Spheroids	Three-dimensional cell aggregates measuring 50-500 µm	Drug testing, regenerative technologies, study of physiological vessel growth, study of angiogenesis in tumor model	The ability of cellular differentiation and interaction with the extracellular matrix, high relevance of models, versatility and cost-effectiveness of models	Limited lifespan, difficulty in size control, uneven diffusion of nutrients	1-2 weeks
Werschler et al[36], 2024	Organelles	Self-organizing 3D structure from stem cells that mimics tissue architecture	Study of cardiotoxicity of drugs, modeling of ischemia, replacement of animal testing, personalization of medicine	High physiological relevance, personalization	Limited size, technical complexity, significant variations from batch to batch, short service life	2-4 weeks
Chen et al[34], 2017	Microfluidics	Technology for working with liquids in channels the size of which is tens of micrometers	Study of thrombosis, atherosclerosis, and the effects of drugs on blood vessels	Reproduction of hemodynamics, increasing structural homogeneity, integration of sensors	High cost, complex production, limited scalability	2-6 weeks
Paek et al[39], 2019	Organ-on-a-chip	Microfluidic system combining cardiomyocytes and endothelial cells	Study of cardiotoxicity of drugs, modeling of ischemia, replacement of animal testing, personalization of medicine	Modeling the interaction of the heart and blood vessels	High cost and complexity of production	4-8 weeks
Jang et al[54], 2019	iPSC models	Using induced pluripotent stem cells to generate endothelial cells	Research of hereditary angiopathies, personalized medicine, determination of cardiotoxicity of drugs	The ability to use patient cells without requiring embryonic stem cells	Difficulty of standardization, high cost and time of implementation, genetic instability	6-10 weeks

At this stage, there is no ideal model of in vitro angiogenesis. Many existing models do not take into account hemodynamics and mechanical stimuli, which reduces their relevance. This problem can be solved by combining methods and creating complex models from 3D models (sandwich, spheroids, organoids, cell layers) with elements of microfluidics and bioprinting. Natural Matrigel can have variability from batch to batch. The problem is solved by using synthetic hydrogels and a constant composition. Also, most models focus on one stage of angiogenesis, ignoring systemic interactions, which can be solved by creating multi-organ chips. AI analysis and machine learning will allow processing large volumes of data from complex experiments and designing optimal vascular architecture to create complex models.

This review provides analysis of in vitro models for studying angiogenesis and highlights their indispensable role in cardiovascular research. From simple 2D assays to advanced 3D microfluidic platforms, each model offers a unique balance between physiological relevance, analytical capability, and cost-effectiveness. No single model is universally optimal; the choice should be aligned with the specific research objective — whether high-throughput screening or mechanistic investigation under biomimetic conditions. Microfluidic organ-on-a-chip platforms are currently emerging as the most promising systems, as they uniquely replicate critical aspects of the vascular niche, including shear stress, multicellular co-cultures, and gradient-driven morphogenesis. Despite substantial progress, there remain major gaps in our ability to accurately model human angiogenesis in vitro.

The key limitations include: Lack of immune integration. Most current models do not incorporate immune cells such as macrophages, which play a critical role in both physiological and pathological angiogenesis, including destabilization of atherosclerotic plaques. Absence of disease-specific microenvironments. There is an urgent need for models that recapitulate pathological features such as oxidative stress, lipid accumulation, and ECM remodeling characteristic of diabetes, atherosclerosis, and other vascular diseases. Insufficient validation and translational alignment. A major challenge is rigorous quantitative validation of complex in vitro models to ensure predictive accuracy for human physiology and therapeutic outcomes.

Looking ahead, several future directions appear particularly promising: Next-generation complexity. The field is moving toward multi-organ platforms (e.g., coupling HoC with vascular constructs) to study systemic interactions, and toward patient-specific models derived from iPSCs to reflect genetic variability. Advanced data acquisition and AI integration. The complexity of 3D systems requires moving beyond conventional microscopy. Integration of high-resolution imaging, automated computational analysis, and AI is essential for objective, high-throughput quantification of vascular networks and for prediction of efficacy and toxicity. Standardization and scalability. Broad implementation of advanced systems will require standardization of protocols, materials, and analytical endpoints, as well as development of more accessible and cost-effective platforms.

CONCLUSION

As the search for an ideal in vitro model of human angiogenesis continues, strategic integration of existing approaches with emerging technologies will be essential. The overarching objective remains the development of reliable human-specific test systems that not only improve mechanistic understanding of cardiovascular disease, but also accelerate the discovery of novel pro- and anti-angiogenic therapies.