Various drug testing platforms are available for patient-specific tumor models, including in vitro 2D systems, 3D systems, organ-on-a-chip systems, and in vivo animal models. Each model has its own advantages and drawbacks8. Previous studies have extensively investigated the molecular mechanisms underlying tumor growth and dissemination through two-dimensional culture models. However, it has been repeatedly demonstrated that these models fail to accurately replicate the complexities of tumor expansion and the structural features of the tumor microenvironment. Consequently, these methods frequently yield inadequate predictions regarding drug responses9. Researchers are dedicated to developing alternative in vitro models for human cancers. In response to this challenge, traditional subcutaneous implants are no longer available because they cannot replicate the growth of organ tissues or the differences in responses to actual treatments10. Instead, human tumor xenografts, known as patient-derived xenografts (PDXs), are implanted into "in situ" organ sites of origin in mice11–14. In vivo in situ cancer models offer a more accurate representation of tumor growth and metastasis. However, there are still challenges in assessing how changes in the microenvironment over time affect cancer cell behavior, and the organ microenvironment may not perfectly mimic that of humans. Moreover, these models necessitate a longer cultivation period, and thus, cannot provide timely information crucial for precise patient treatment, particularly in the context of metastatic brain tumors.
The PDTSs utilized in this study hold significant promise for advancing personalized cancer medicine. These 3D cultures, which are derived from a patient's own tumor tissue, better capture the heterogeneity and complexity of the original tumor than traditional 2D cell cultures. PDTSs not only serve as valuable tools for basic cancer research but also offer capabilities for drug screening and predicting treatment responses15. They boast several advantages over 2D systems, including greater physiological relevance, suitability for clinical tests, and support for high-throughput screening4,6,16,17.
In our previous study18, we explored the potential of patient-derived organoids (PDOs) as a platform for precision treatment of malignant brain tumors. Our prospective clinical investigation demonstrated that PDOs derived from both primary and metastatic brain tumors could effectively guide treatment selection, thereby enhancing patient survival rates. However, the success rate of organoid culture remains at only 50%, with a culture time of approximately 2–3 weeks. Additionally, quantifying total cell numbers in PDOs presents challenges, making it difficult to accurately assess drug efficacy. These limitations hinder the clinical application of PDOs.
The PDTS model, resembling a 3D system, features a unique biomimetic environment that incorporates a brain tumor spheroid and a functional endothelial cell layer with good cell–cell interactions, mimicking the blood–brain barrier (BBB) in vivo. This allows for a more accurate simulation of drug delivery within the human vascular system. Additionally, the model facilitates comprehensive recreation of cellular interactions, including those between cells and the extracellular matrix (ECM), as well as variations in chemical and physical gradients within the cellular microenvironment. Importantly, it addresses the limitations of traditional 3D in vitro models, such as the absence of a functional vascular network for nutrient and gas transportation and the removal of toxic byproducts generated by cells. The PDTS model offers several advantages. It can provide drug test results to doctors within three weeks, enabling real-time decision-making. Moreover, our platform allows simultaneous detection of multiple drugs and quantitative testing of different drug concentrations, enhancing the accuracy of reports. Additionally, by culturing and storing tumor cells, our system ensures reproducibility and facilitates the retesting of other drugs, providing further clinical insights.
However, our model has certain limitations. Particularly for metastatic tumors, variations in growth and differentiation often result in differences across different organs of the human body. Furthermore, growth patterns and drug responses can significantly differ based on the unique microenvironment of each organ. Consequently, when using the same model, the predictive value of drug tests may vary between the brain and the organs of origin. Additionally, our PDTS model does not include a mechanism for studying angiogenesis, which prevents our testing of angiogenesis inhibitors such as bevacizumab. Moreover, our model does not incorporate the immune system, and the immune system is a vital component for evaluating the efficacy of immunotherapy, which is a commonly used treatment approach today. Another factor that could impact the accuracy of the PDTS model is the heterogeneity of clinical outcomes due to patients receiving multimodal management, including radiotherapy, targeted therapy, chemotherapy, immunotherapy, or cell therapy. In our study, two patients received cell therapy, one received hormone therapy, and one underwent oophorectomy for breast cancer, potentially influencing the outcomes of medical treatments and subsequently affecting the accuracy of testing. In this study, we employed a novel PDTS model to replicate the microenvironment of brain tumors, aiming to investigate the correlation between patient clinical outcomes and findings from ex vivo PDTSs. To achieve this goal, we utilized bovine adipose-derived extracellular matrix (dECM) as a supportive scaffold for brain tumor cells and cocultured a layer of endothelial cells to simulate the blood‒brain barrier (BBB) and for drug delivery. A coclinical trial involving 17 patients with a total of 18 metastatic brain tumors was conducted to validate the results obtained from the ex vivo PDTS models. The PDTS model demonstrated an accuracy rate of 73% for anticancer drug testing in metastatic brain tumors, with a specificity of 75% and a sensitivity of 71%. This model has the potential to become a standard in vitro model for clinical applications, as it can effectively simulate activities within a patient’s body that resemble the microenvironment and organ systems, including mechanical and physiological responses. Additionally, it offers a cost-effective and high-throughput approach, thereby contributing to advancements in cancer treatment. Given the frequently severe clinical condition of patients with brain metastases, it is crucial to note that recruiting participants for this study has been challenging. To advance this clinical study, a larger sample size and longer follow-up periods are needed.