The blood-brain barrier (BBB) is a highly selective barrier protecting the brain's parenchyma. Its intricate structure, composed of endothelial cells, pericytes, astrocytes, microglia, and neurons (the neurovascular unit or NVU), along with the extracellular matrix (basal lamina or BL), ensures nutrient supply while preventing pathogen and harmful molecule entry. BBB dysfunction is implicated in various CNS disorders, including Alzheimer's, Parkinson's, amyotrophic lateral sclerosis, and even COVID-19. This dysfunction presents a major challenge for drug delivery, as many therapeutics cannot effectively penetrate the BBB. Therefore, the development of accurate in vitro models that can predict drug permeability across the BBB is crucial for accelerating the development of effective therapies for these conditions. Traditional animal models, while widely used, suffer from interspecies differences that can lead to inaccurate predictions and high costs. In vitro 2D static models using animal cells also have limitations in replicating shear stress and the complex cellular interactions of the NVU. These limitations necessitate a move towards more sophisticated, biomimetic in vitro models to address this critical need for accurate and cost-effective preclinical testing.

The literature extensively covers the challenges in drug delivery to the CNS due to the BBB's highly selective permeability. Molecules can cross the BBB via paracellular or transcellular transport. Paracellular transport is limited by tight junctions between endothelial cells, while transcellular transport faces challenges due to efflux transporters like P-glycoprotein, BCRP, and MRP2, which actively pump drugs out of endothelial cells. Other NVU cells also express similar transporters. Strategies for enhancing drug delivery include temporarily disrupting BBB integrity (though this is invasive and risky) or modifying drugs to enhance lipid solubility. Smart drug delivery systems, such as nanoparticles, are also being explored, focusing on ligand decoration to promote receptor-mediated transcytosis. However, the lack of realistic in vitro models is a significant obstacle for clinical translation. Existing in vitro BBB models, including Transwell-based models, have limitations in mimicking in vivo conditions, such as the lack of shear stress, incomplete NVU representation, and the absence of a physiologically accurate basal lamina. The review extensively cites studies highlighting these limitations and the need for improved biomimetic models.

This review systematically analyzes the evolution of in vitro BBB models, starting from traditional animal models and 2D static cultures. The limitations of these approaches, such as the lack of shear stress and the incomplete representation of the NVU, are discussed in detail. The transition to Transwell-based models incorporating multiple cell types is examined, highlighting the improvements in barrier integrity as measured by trans-epithelial electrical resistance (TEER). The incorporation of shear stress using microfluidic devices is discussed as a further improvement. The critical role of the basal lamina (BL) in BBB integrity and the different strategies for mimicking its composition and structure using porous membranes (polycarbonate, polyester, PET, etc.) or hydrogels (collagen, fibrin, Matrigel) are analyzed. Various studies employing organ-on-chip (OOC) technologies are reviewed, focusing on the advantages of 3D cell culture, dynamic flow conditions, and the potential for integrating biosensors. Specific OOC models used for drug evaluation and the analysis of brain pathologies (glioblastoma, Alzheimer's disease, virus infection, ischemic stroke) are detailed, highlighting their strengths and limitations. The minimum requirements for a truly biomimetic BBB model—a tri-culture of human endothelial cells, pericytes, and astrocytes within a hydrogel mimicking the basal lamina—are established. The review also explores the utility of incorporating neurons in the model to assess drug neurotoxicity and the use of multi-organ chips to study drug metabolism before BBB interaction.

The review's key findings center on the limitations of existing BBB models and the advantages of OOC technology for creating biomimetic in vitro systems. Animal models, while historically used, suffer from interspecies differences affecting the accuracy of drug testing and ethical concerns. 2D static in vitro models fail to capture crucial aspects such as shear stress and complex NVU interactions. Transwell systems show some improvement but lack flow and a fully representative BL. OOC technology overcomes many of these limitations by allowing 3D cell cultures with dynamic flow, enabling a more realistic mimicry of the in vivo environment. The review emphasizes that for a model to be truly biomimetic it needs at least a tri-culture of human endothelial cells, pericytes, and astrocytes, with a hydrogel mimicking the BL to ensure proper cell-cell contact. The inclusion of neurons is also highlighted as crucial for evaluating neurotoxicity. The review presents numerous examples of OOC BBB models applied to drug testing and the investigation of various brain pathologies, with a detailed discussion of the strengths and weaknesses of each model. TEER, while useful, is not sufficient as a sole indicator of model validity, and additional measures are essential. The study concludes that a biomimetic OOC model is a substantial step towards replacing animal models, accelerating drug discovery, and enhancing preclinical data.

This review strongly advocates for the adoption of biomimetic OOC BBB models as a superior alternative to traditional methods. The findings directly address the limitations of existing models and present a clear path toward improving the accuracy and efficiency of preclinical drug development. The shift from animal models and simplistic in vitro systems to sophisticated OOC models is justified by the significant limitations of the former, particularly interspecies variability and the inability to accurately capture the complex dynamics of the BBB. The emphasis on specific criteria—a tri-culture of human cells within a hydrogel—provides a practical framework for researchers designing new models. The incorporation of biosensors for real-time monitoring further enhances the reliability and potential of this technology. The discussion of various applications, from drug permeability studies to disease modeling, highlights the versatility and potential impact of OOC technology in accelerating drug discovery and personalized medicine approaches for brain-related diseases.

This review establishes the need for biomimetic in vitro BBB models and highlights the advantages of OOC technology. The minimum requirements for biomimicry—a human tri-culture (endothelial cells, pericytes, and astrocytes) within a hydrogel simulating the BL—are proposed. OOC models provide dynamic flow, allowing for the study of shear stress effects. Integrating neurons and employing multi-organ chips for comprehensive drug evaluation are suggested for future advancements. OOC technology offers significant potential to reduce reliance on animal models, accelerating drug discovery and clinical translation.

While this review extensively covers the advantages of OOC BBB models, it should be noted that the complexity of these systems also presents limitations. The cost and technical expertise needed to create and maintain OOC models might be higher than traditional methods. The long-term stability and reproducibility of these complex systems also require further investigation. Furthermore, while OOC models strive for biomimicry, they still represent a simplification of the intricate in vivo environment. The perfect in vitro simulation of the human BBB remains a challenging goal.