Organoids in cancer therapy: translational applications and clinical promise Authors: Mingzhen Bai, Maoyun Liu, Sheng Wang, Ming Ran, Jingyuan Ye, Yuqin Xie, Cheng Zhang, Yi Wang, Jiale Lu, Fengsheng Dai, Dewei Li, Hui Li Published: 2026 ABSTRACT AbstractCancer remains a leading cause of mortality worldwide, with its management challenged by profound biological complexity and substantial intratumoral and interpatient heterogeneity. Conventional preclinical tumor models fail to recapitulate key features of the human tumor microenvironment, limiting their translational relevance. Tumor organoids are three-dimensional (3D) culture systems that retain patient-specific tissue architecture and heterogeneity, and have emerged as powerful platforms for translational cancer therapy research. Recent advances in organoid engineering, including microfluidic organoid-on-chip systems, 3D bioprinting, and next-generation biomaterials, have markedly improved physiological relevance, experimental controllability, and scalability. These innovations enable higher-throughput drug screening and more informative functional profiling to support individualized therapeutic decision-making. In parallel, organoid-based predictive modeling frameworks are creating new opportunities to rationally design precision treatment regimens and to elucidate mechanisms underlying therapeutic response and acquired resistance. Importantly, integrating organoid-derived functional readouts with molecular and clinical data is accelerating bench-to-bedside translation and strengthening the evidence base for personalized oncology. Here, we systematically summarize the multidimensional applications of organoids in translational cancer therapy, critically discuss their advantages and current limitations, and outline key directions for future development, with the aim of facilitating the implementation and optimization of precision oncology.Keywords: Organoid, Precision oncology, Tumor microenvironment, Drug screening, Organoid engineering, Translational cancer therapy Introduction Cancer persists as a predominant cause of mortality globally. Advances in molecular subtyping and precision medicine have broadened the clinical use of targeted therapies and immunotherapies. Nonetheless, clinical benefit is often limited by biological complexity and pronounced intratumoral as well as interpatient heterogeneity [1, 2]. Translational progress is further constrained by preclinical models that incompletely capture key features of human tumors. Conventional two-dimensional (2D) cultures lack physiologically relevant three-dimensional cell–cell and cell–matrix interactions and therefore inadequately model spatial organization, cellular diversity, and microenvironment-driven plasticity, contributing to discrepancies between experimental findings and clinical outcomes [3–5]. Patient-derived xenografts (PDXs) preserve elements of tissue architecture and therapeutic response and are widely used for drug validation and resistance studies. Nevertheless, interspecies differences, ethical constraints, long establishment times, and high costs limit scalability and experimental control [6, 7]. Collectively, these limitations underscore the need for human-relevant in vitro models that preserve patient-specific tumor features while enabling mechanistic interrogation and higher-throughput testing. Organoid technology offers a powerful alternative. Organoids are self-organizing three-dimensional (3D) cultures derived from embryonic or induced pluripotent stem cells, or from patient tumor cells [8, 9]. While they recapitulate key aspects of organ architecture, faithful modeling of the tumor microenvironment (TME) typically requires explicit engineering. Their conceptual origins trace to early observations of cellular self-organization, later enabled by extracellular matrix (ECM)–based 3D culture systems [10–13]. The landmark development of the intestinal organoid model by Hans Clevers’ group in 2009 laid the foundation for organoid-based disease modeling [10]. Since the first colorectal cancer organoids were reported [14], this technology has been rapidly extended to a wide range of malignancies, including prostate [15], pancreatic [16], glioblastoma [17], ovarian [18], hepatocellular [19], renal [20], bladder [21], breast [22] and gastric cancers [23]. Patient-derived organoids (PDOs) retain clinically relevant tumor architecture and heterogeneity, enabling functional assessment of drug responses, investigation of resistance mechanisms, evaluation of immunotherapy efficacy, and identification of predictive biomarkers (Fig. 1). Timeline of organoid culture technology development More recently, organoid research has progressed toward engineering optimization and clinical translation. Explicit engineering strategies, such as integration with microfluidic organoid-on-chip platforms, genome editing, and co-culture with immune components, enable the reconstruction and modeling of key TME features in vitro [24–27]. As these approaches mature, organoid models are evolving from static cultures into dynamic, tissue-like ecosystems [28]. Concurrently, combining organoid-based functional assays with multi-omics and artificial intelligence is enabling predictive modeling of drug responses and accelerating development of individualized therapeutic strategies [29]. This review adopts an engineering–prediction–translation framework to synthesize recent advances in organoid-based cancer research. We summarize engineering strategies that enhance physiological relevance and scalability, describe predictive approaches for evaluating therapeutic sensitivity, and discuss emerging avenues for clinical integration. Finally, we outline current challenges and future directions for advancing organoid-enabled precision oncology. Organoid construction and engineered evolution With continuous advancements in bioengineering, materials science, and microenvironment simulation technologies, organoid research has entered a phase of rapid evolution and systemic restructuring. From the initial three-dimensional culture systems reliant on Matrigel, organoid technology has evolved from simple cell self-organization models into controllable engineered microtissue platforms, incorporating complex models such as biodegradable scaffolds, microfluidic chips, 3D bioprinting, and gene editing. These technological innovations have significantly enhanced organoid reproducibility, structural stability, and physiological relevance, enabling increasingly widespread and precise applications in tumor biology research, drug screening, and personalized therapy (Fig. 2). Engineering strategies enabling translational organoid platforms. (a) Optimization of 3D culture matrices and biomaterials (GelMA, Matrigel, dECM hydrogels) to improve structural fidelity. (b) Integration of diverse microfluidic systems to simulate dynamic perfusion, gradient generation, and multi-organ interactions. (c) Application of 3D bioprinting technologies with advanced bioinks and scaffold designs for spatially organized, vascularized organoid construction. (d) Genetic engineering techniques for functional genomics and personalized modeling Optimization of three-dimensional culture matrices and biomaterials In organoid culture systems, the selection and optimization of 3D matrices and biomaterials are fundamental for maintaining organoid architecture, functionality, and long-term viability. Beyond providing structural support, the matrix serves as a key determinant of cell–matrix signaling, tissue organization, and morphogenetic behavior, thereby shaping how faithfully organoids recapitulate native human tissues. As the field advances, matrix engineering is shifting from static support systems toward more programmable platforms with improved reproducibility, scalability, and translational relevance. Matrigel has been the gold standard for organoid culture due to its rich bioactive components, which support cell adhesion, proliferation, and differentiation. It has been extensively used across various cancer types, providing valuable insights into tumor biology [30]. However, its use is hindered by several limitations, including batch-to-batch variability, undefined composition, and poor control over mechanical properties such as stiffness [31, 32]. To partially overcome these limitations, naturally derived alternatives such as Gelatin methacrylate (GelMA) have gained prominence. GelMA combines cell-adhesive biological motifs with light-responsive crosslinking, thereby enabling tunable stiffness and structural control [33–35]. Matrix stiffening can modulate genes involved in proliferation and extracellular matrix remodeling, thereby influencing cell adhesion and migration [35]. However, mechanical tunability alone is not sufficient, as the absence of critical biochemical cues may alter organoid polarity and growth behavior, emphasizing the need to integrate both mechanical and biochemical design principles in engineered matrices. Decellularized ECM (dECM) hydrogels extend the advantages of natural matrices by preserving tissue-specific biochemical cues while reducing immunogenicity [36, 37]. For example, kidney-derived dECM hydrogels promote the differentiation of human pluripotent stem cells into kidney organoids and support the formation of endogenous vascular structures, highlighting the role of native ECM in functional maturation [38]. Tissue-specific dECM can also enhance disease phenotype fidelity. Song et al. demonstrated that a decellularized uterine cervical ECM hydrogel derived from patient cervical tissue supported the growth of cervical squamous cell carcinoma organoids, which recapitulated primary tumor morphology, transcriptomic profiles, and drug responses [39]. Together, these findings position dECM hydrogels as an important intermediate between undefined natural matrices and fully synthetic scaffolds by preserving tissue-specific biochemical identity while improving biological relevance. Synthetic matrices further advance matrix engineering by enabling defined and independent control of biochemical signals, stiffness, and degradation kinetics. Notably, Gjorevski et al. demonstrated that modular synthetic hydrogels with independently tunable mechanical and biochemical properties can directly regulate intestinal stem cell expansion, differentiation, and organoid formation, thereby defining key extracellular matrix parameters governing organoid fate [40]. Complementing this, Cruz-Acuña et al. showed that a fully defined polyethylene glycol (PEG)-based synthetic hydrogel supports robust and highly reproducible growth of human intestinal organoids without reliance on tumor-derived ECM, further highlighting the translational potential of synthetic matrices [41]. More broadly, synthetic matrices, such as PEG-based hydrogels and alginate composites enable precise control over matrix composition and mechanics [42, 43]. By altering crosslinking density and material composition, these systems can approximate the elastic modulus of distinct tissues and thereby influence tumor cell behavior and signal transduction. For example, PEG hydrogels modulate colony morphology and epidermal growth factor receptor signaling in colorectal cancer cells, illustrating the impact of matrix stiffness and biochemical composition on tumor cell phenotype [44]. Beyond stiffness alone, dynamic mechanical properties such as viscoelasticity, stress relaxation, and reversible crosslinking are increasingly recognized as critical regulators of organoid morphogenesis and functional maintenance [45, 46]. Collectively, these advances are shifting organoid matrices from passive scaffolds toward dynamically programmable systems with improved mechanistic control, reproducibility, and translational relevance for tumor modeling. Application of microfluidic technology in organoid culture Microfluidic technology has emerged as a powerful engineering platform for enhancing the physiological relevance and experimental controllability of organoid systems. An important early demonstration of organ-on-chip technology was provided by Huh et al., who established a biomimetic lung-on-a-chip model that reconstituted a critical alveolar–capillary interface and demonstrated organ-level responses under dynamic mechanical and microfluidic control [47]. Building on this foundation, microfluidic platforms have increasingly been integrated with organoid systems to enable programmable perfusion, biochemical gradients, and real-time monitoring in vitro [48]. In contrast to conventional static cultures, which inadequately reproduce spatiotemporal gradients of oxygen, nutrients, metabolites, and drugs, microfluidic platforms allow controlled perfusion and gradient formation, thereby supporting a more physiologically relevant tumor niches and treatment conditions [49, 50]. Functionally, integrating organoids with microfluidics introduces both mechanical constraints and controlled cellular microenvironments, thereby improving the fidelity of drug screening and therapeutic response assessment. In pancreatic ductal adenocarcinoma (PDAC), perfusion-based chip systems replicate key tumor features such as elevated interstitial pressure and hypoxia, and have been used to assess treatment sensitivity [51]. An HCC-PDO-on-chip model incorporating mesenchymal stem cells (MSCs) and immune cells reconstructed an immunosuppressive microenvironment in vitro, enhanced PDO formation efficiency, and supported high-throughput screening for anti–PD-L1 response evaluation [52]. Similarly, tumor-on-a-chip platforms integrating pancreatic organoids, astrocytes, and macrophages have recapitulated matrix–immune interactions that regulate chemotherapy sensitivity, providing a platform for personalized drug evaluation [53]. Together, these advancements demonstrate the potential of microfluidic organoid systems for modeling tumor microenvironments and drug responses with increasing precision, making them increasingly relevant for personalized cancer therapy. To address clinical demands for rapid assessment, droplet-based microfluidics enables the high-throughput generation of micro-organoid spheres from minimal biopsy material. Ding et al. leveraged emulsion microfluidics to produce thousands of uniform PDOs within days, facilitating drug sensitivity testing and immunotherapy response profiling while significantly reducing culture time and enhancing clinical applicability [54]. Similar strategies have also been applied to biopsy-derived pancreatic cancer organoids, demonstrating the feasibility of drug testing from limited patient material under controlled microfluidic conditions [55]. Beyond droplet-based rapid organoid generation, automated high-throughput microfluidic platforms have further enabled dynamic and combinatorial drug screening in tumor organoids under controlled conditions [56]. More recently, microfluidic systems based on single-cell-derived tumor organoid arrays have further advanced the development of personalized screening by addressing heterogeneity-mediated chemoresistance in colorectal cancer [57]. Collectively, these studies suggest that microfluidics is evolving not only as a tool for microenvironmental engineering, but also as a practical platform for rapid and potentially clinically actionable organoid-based testing. From an engineering standpoint, microfluidic devices fabricated from biocompatible materials such as polydimethylsiloxane enable multi-layered channel configurations, gradient generators, and precise control over flow rates, drug concentrations, and exposure durations. These capabilities effectively mimic in vivo pharmacokinetics, support long-term culture, and alleviate cellular stress responses, including endoplasmic reticulum stress [58]. Integration with electrochemical and optical sensors further allows real-time monitoring of metabolic activity, oxygen levels, and drug responses. For instance, Ruan et al. developed a vascularized tumor organoid chip coupled to a real-time ELISA module, enabling continuous monitoring of tumor biomarkers and therapeutic efficacy [59]. Microfluidic technology further enables the development of multi-organoid-on-chip systems, extending organoid modeling from local tissue niches to inter-organ physiological interactions. Early work by Skardal and colleagues helped define this direction conceptually and experimentally, including multi-organoid body-on-a-chip platforms for drug screening and a three-tissue organ-on-a-chip system demonstrating drug responses dependent on inter-tissue communication [60, 61]. These advances broaden the application of organoid platforms in evaluating systemic pharmacokinetics and drug toxicity. A vascularized PDO-on-chip model has been shown to dynamically recapitulate tumor angiogenesis and cell migration, offering a valuable tool for investigating metastasis and anti-angiogenic therapies in vitro [62]. Additionally, incorporation of decellularized small intestinal submucosal matrix into microfluidic platforms facilitates the automated and uniform generation of colorectal cancer PDO microspheres without reliance on Matrigel, significantly enhancing model standardization, reproducibility, and scalability [63]. Collectively, these advances are shifting organoid cultures from static experimental systems toward increasingly programmable and clinically relevant platforms with translational potential. Bioprinting technology facilitates spatial construction of organoids 3D bioprinting, which integrates biofabrication and tissue engineering, has emerged as a powerful tool for reconstructing the spatial architecture and microenvironmental complexity of organoids. Compared with conventional culture methods, bioprinting offers precise control over key parameters such as printing resolution, layer thickness, cell density, and crosslinking conditions. These capabilities enhance morphological and mechanical fidelity to in vivo tumors while improving reproducibility and standardization [64, 65]. The performance of bioprinted organoids depends largely on the printability, biocompatibility, and biofunctionality of the bioinks used. Recent advances in composite and functionalized bioinks have markedly expanded their applicability. For example, an embedded bioprinting approach combining low‑concentration collagen I bioink with a sericin‑supported bath enabled co‑culture of breast cancer cells and cancer‑associated fibroblasts (CAFs). This system reproduced hallmark tumor features, including epithelial–mesenchymal transition (EMT) and angiogenesis, thereby supporting the construction of physiologically relevant breast cancer models [66]. Likewise, a composite bioink formed from porcine adipose-derived extracellular matrix (adECM) and GelMA enabled high-throughput fabrication of colorectal cancer organoids. These constructs exhibited superior structural stability and drug responsiveness compared with traditional cultures, providing an efficient biomimetic platform for standardized drug screening [67]. In parallel, microgel-based assembly and reversible crosslinking strategies improve molding precision and confer tunable viscoelasticity, facilitating signal integration and fine control of local microenvironments [68, 69]. The translational value of bioprinting is reflected in both mechanistic research and individualized therapy. Mechanistically, bioprinted chimeric breast cancer organoids have demonstrated that normal epithelial cells can modulate tumor metabolism and fate through mitochondrial transfer, offering insight into tumor microenvironment–mediated metabolic reprogramming [70]. Clinically, patient-specific bioinks—such as lung tissue–derived decellularized ECM—enable fabrication of vascularized lung cancer organoids that preserve patient-specific genomic alterations and spatial heterogeneity [71]. Extrusion-based automated platforms further allow rapid production of dimensionally uniform tumor organoids for drug sensitivity testing and large-scale screening [72, 73]. Despite rapid progress, several challenges remain. Trade-offs persist between printability and physiological functionality, stable vascularization and perfusion remain difficult to achieve, and standardized quality control for complex multicellular constructs is lacking. Nonetheless, recent progress in high-resolution printing, dynamically crosslinkable biomaterials, and intelligent bioinks is expected to improve the reproducibility of spatially engineered organoid models and facilitate their translational validation in clinical and pharmaceutical contexts. Genetic engineering Genetic engineering, particularly CRISPR–Cas9 technology, has transformed tumor organoids into platforms for causal dissection of tumor initiation, progression, and therapeutic vulnerability. This technology enables precise gene perturbation, functional screening, and mechanistic interrogation, while maintaining human genetic backgrounds and three-dimensional tissue architecture [74, 75]. CRISPR-based approaches have been applied to model tumorigenesis, study metastasis, and identify candidate therapeutic targets in organoid-derived tumors. For instance, He et al. performed candidate tumor suppressor screens in gastric mouse organoid–derived tumors, identifying genes that constrain tumorigenesis and actively shape the TME [76]. Extending this paradigm to PDOs enables interrogation of clinically relevant processes, including metastatic progression. In colorectal cancer, a genome-wide CRISPR screen conducted in a liver metastasis model established by intrahepatic injection identified ANKRD42 as a critical regulator of metastatic colonization [77]. Beyond tumor initiation, CRISPR-enabled forward genetic modeling further allows reconstruction of multi-step oncogenic pathways in human-relevant systems. Lo et al. deleted ARID1A in human gastric organoids and combined this alteration with TP53 deficiency, thereby recapitulating the key morphological and transcriptomic features of ARID1A-mutant gastric cancer and identifying the FOXM1–BIRC5 axis as a potential therapeutic vulnerability through high-throughput screening [78]. Similarly, Kawasaki et al. demonstrated how genetic inactivation and transcriptional reprogramming collaborate in gastroenteropancreatic neuroendocrine carcinoma formation [79]. In precancerous models, loss of CTTN suppressed adenocarcinoma development, highlighting membrane dynamics and exosome secretion as critical regulators of tumor initiation [80]. Together, these studies demonstrate that CRISPR-based engineering enables organoids to move beyond descriptive disease modeling toward causal dissection of tumor initiation, lineage plasticity, and metastatic progression in human-relevant systems. The ability to model genetic complexity has been enhanced through integration with transposon-based systems. Bian et al. introduced MYC amplification and oncogenic mutations in a cerebral organoid model, generating glioblastoma-like tumors and enabling functional validation of tumor initiation and evolution [81]. Translationally, pooled CRISPR screening has been adapted to PDOs. In breast cancer PDOs, kinase-focused screening identified key survival dependencies and suggested synergistic inhibition of EGFR and FGFR1 as a therapeutic strategy [82]. In parallel, precise genome editing and base-editing approaches further permit construction of isogenic organoid pairs that differ only at specific driver loci. Millen et al. edited hotspot mutations, including PIK3CA E545K, directly linking mutation status with targeted-therapy response and supporting genotype-guided patient stratification [83]. These advances further establish genetically engineered organoids as scalable functional genomics platforms that can simultaneously support mechanistic interrogation, vulnerability discovery, and genotype-guided therapeutic stratification. Overall, genetic engineering has evolved tumor organoids from passive disease models to scalable platforms for causal modeling, functional screening, and therapeutic stratification. However, editing efficiency varies across tumor types and organoid systems, while single-cell dissociation and clonal expansion may introduce selection bias and reduce intratumoral heterogeneity [84]. Off-target effects and the complexity of multiplex perturbations can further complicate phenotype interpretation. Moreover, unless combined with stromal or immune co-culture systems, genetically engineered PDOs mainly inform tumor-intrinsic mechanisms. Greater standardization of editing workflows and quality control will therefore be essential for improving reproducibility and translational relevance. Tumor organoids for therapeutic development and response prediction A key challenge in oncology remains the discrepancy between preclinical model performance and real-world clinical outcomes. Traditional 2D cell lines fail to retain tumor spatial architecture and heterogeneity, while PDXs are limited by long establishment times, low throughput, and species-specific microenvironmental differences [85]. In contrast, PDOs provide a more scalable and physiologically relevant platform for therapeutic development by preserving key tumor-intrinsic architectural and genetic features of the original lesions. However, standard PDOs primarily capture tumor cell-autonomous phenotypes, and their predictive value for therapies that depend strongly on stromal, vascular, or immune components still requires further engineering. Within these boundaries, PDOs have emerged as versatile platforms for drug-response profiling, resistance mechanism analysis, biomarker discovery, and optimization of combination regimens, thereby supporting precision oncology (Fig. 3). Workflow of PDO construction and application for personalized oncology. Tumor samples are obtained via surgical resection or biopsy. Following tissue dissociation, PDOs are established using 3D culture system. High-throughput drug screening is then conducted on PDOs using a panel of therapeutic agents. Drug sensitivity is evaluated based on viability assays, such as ATP-based assays, to generate individualized response profiles. Personalized treatment regimens are subsequently formulated based on the organoid drug response spectrum. Selected therapeutic strategies undergo clinical validation to inform precision oncology–guided decision-making Drug screening and sensitivity profiling PDO-based drug screening provides a functional framework for individualized treatment stratification. Through high-throughput culture systems and quantitative phenotypic assays, PDOs can generate patient-specific drug response profiles within days. While PDO-based assays have demonstrated notable concordance with clinical drug responses in multiple tumor types, such as pancreatic ductal adenocarcinoma, colorectal cancer, and non-small cell lung cancer, the overall predictive accuracy and clinical applicability vary across different cancer types and treatment contexts [86–89]. Overall, however, predictive performance remains context dependent and varies across cancer types, therapeutic classes, and assay configurations. Multiple studies support the utility of PDOs as reliable preclinical drug screening models. In bladder cancer, PDOs preserve the histological and immunohistochemical characteristics of the original tumor, and the in vitro drug responses correlate with patient outcomes [90]. Malignant ascites-derived organoids from ovarian cancer patients can be generated rapidly and assessed via transcriptomic and proliferative markers, enabling timely drug testing to inform personalized treatment decisions [91]. In gastric cancer, large-scale studies demonstrate that PDOs reproduce inter-patient heterogeneity in response to standardized chemotherapeutics. Drug sensitivity results from PDOs show strong agreement with both xenograft models and clinical cohorts, supporting their predictive value for chemotherapy selection [92]. Similarly, organoids derived from malignant ascites of gastrointestinal cancers retain tumor histopathology, with drug response matching clinical outcomes at an accuracy rate of 78.57–85.71% [93]. In colorectal cancer (CRC), multiple studies confirm that PDO-based drug sensitivity scores correlate with progression-free survival and tumor shrinkage. A higher in vitro “organoid score” was associated with shorter clinical benefit duration following standard therapy [94]. Integration of PDO drug response data with clinicopathological models enhances prognostic stratification. In a postoperative stage IV CRC cohort, in vitro resistance to chemotherapeutics was significantly associated with inferior outcomes, and PDO-based drug testing emerged as an independent predictor in multivariate analysis [95]. In a prospective study of 232 metastatic CRC patients, a 75% PDO success rate was achieved. PDO-based responses to 5-FU and oxaliplatin strongly correlated with tumor shrinkage, progression-free survival, and overall survival (AUC: 0.78–0.88), providing robust evidence for using PDOs as functional companion diagnostics [96]. More broadly, the diagnostic and predictive value of organoid technology ultimately depends on the robustness of the correlation between organoid-based readouts and actual clinical outcomes. In pancreatic cancer, PDO drug responses have also shown concordance with in vivo tumor growth inhibition in matched PDX models, providing additional support for the translational relevance of organoid-based pharmacotyping [97]. Beyond efficacy prediction, PDOs facilitate rapid screening of new drugs and combinations. For example, PDOs allow sensitivity profiling within one week, supporting neoadjuvant regimen selection [95]. In pancreatic cancer, PDOs help differentiate patient responses to FOLFIRINOX versus gemcitabine–paclitaxel regimens, enabling therapy optimization and toxicity reduction [98]. Similar applications have been reported in hepatocellular carcinoma, glioma, renal cell carcinoma, gallbladder cancer, oral squamous cell carcinoma, cervical cancer, and lung cancer [99–105]. In lung cancer, organoids established from fresh tissue, bronchoalveolar lavage fluid, or pleural effusion and coupled with high-throughput assays revealed heterogeneous drug responses associated with underlying genomic alterations, further supporting their value in functional precision oncology [105, 106]. PDOs can also serve as validation platforms for emerging therapeutic strategies. For instance, low-dose chemotherapy combined with barium titanate nanoparticle-mediated photodynamic therapy exhibited patient-specific efficacy in PDOs, supporting its potential in personalized combination regimens [107]. The growing application of PDOs in clinical trials highlights their increasing role in personalized cancer treatment. As summarized in Table 1, recent clinical trials initiated since 2023 have explored the use of PDOs across various tumor types for drug testing, screening, and model development. Collectively, these studies indicate that PDOs are increasingly being integrated into clinical oncology workflows as functional platforms for therapy selection and translational hypothesis testing, although their implementation still depends on cancer type, assay standardization, and the degree to which clinically relevant microenvironmental features are preserved. Table 1Clinical trials on PDOs in cancerNCT NumberCancer typeStatusApplicationOutcomeMeasurementStart dateNCT03890614Hematologic malignancyRecruitingDrug testingCoherence assessmentSensitivity,Consistency2019-05NCT06692491Rare tumorNot yet recruitingDrug screeningORR, DCR2025-01NCT03952793Prostatic neoplasmTerminatedPDOsestablishmentPDOs successrate2019-12NCT06536725Prostate cancerRecruitingDrug testingORR, PFS2024-06NCT05577689Prostate neoplasmsRecruitingDrug testingSensitivity2023-08NCT07004582Prostate cancerNot yet recruitingPDOs establishmentPDOs successrate2025-10NCT06529549Metastatic castration-resistant prostate cancerRecruitingDrug testingORR, PFS, OS2024-07NCT06155370Gynecologic cancerUnknownPDOsestablishmentPDOs successrate2023-12NCT06929845Hepatocellular carcinomaRecruitingPDOsestablishmentCorrelation, DFS, OS2024-10NCT06355700Hepatocellular carcinomaRecruitingPDOs establishmentDrug testingPDOs successRateCorrelation2023-06NCT05913141Liver cancer/ Metastatic liver cancerRecruitingDrug testingORR, PFS, RFS, OS2026-06NCT06077591Hepatocellular carcinoma/Colorectal cancerRecruitingDrug testingPFS, OS2024-10NCT02436564Cholangiocarcinoma/Hepatocellular carcinoma/Pancreatic neoplasmUnknownPDOs establishmentGenetic concordance2026-07NCT05024734Hepatocellular carcinomaRecruitingDrug ScreensSensitivity2023-02NCT05932836Hepatocellular carcinomaActive not recruitingPDOs establishmentDrug testingPDOs successrateSensitivity2023-03NCT05644743Intrahepatic cholangiocarci-nomaNot yet recruitingDrug testingDrug resistance prediction2023-01NCT05518110Pancreatic cancerRecruitingPDOs establishmentDrug testingPDOs success rateCell survival2023-05NCT07122466Metastatic pancreatic cancerActive not recruitingDrug testingSensitivity2023-05NCT06615830Pancreatic cancerNot yet recruitingDrug screeningSensitivity Correlation2024-10NCT06246630Advanced pancreatic neuroendocrine tumorRecruitingPDOsestablishment, Drug testingORR, CR, PR2024-04NCT06666803Pancreatic ductal adenocarcinomaNot yet recruitingPDOs EstablishmentPDCOs successrate2024-11NCT06519500Gastroentero-pancreatic NeoplasmsRecruitingPDOs establishmentPDCOs successrate2023-03NCT05842187Pancreatic cancer/Gastric cancerUnknownDrug testingDCR, PFS2023-03NCT06813079Pancreatic ductal carcinomaNot yet recruitingDrug testingORR, DCR, PFS, OS2025-02NCT05927298Pancreas cancerRecruitingPDOs establishmentCorrelation2023-03NCT06102824Breast cancerRecruitingDrug testingPFS, OS, ORR2024-10NCT06786663Breast neoplasmsRecruitingDrug testingCorrelation2024-04NCT06702800Breast neoplasmsRecruitingDrug screeningCR, PR2024-06NCT06468124 Breast neoplasmsRecruiting Drug testingRR, PFS2025-05NCT06084676Breast cancerTerminatedDrug testingORR2024-01NCT07047495 Breast adenocarcinoma/prostate cancerRecruiting PDOs establishmentRadiotherapy validation2026-07NCT06438055 Refractory breast cancerNot yet recruitingDrug testingORR,2026-06NCT05404321 Triple negative breast cancerRecruiting PDOs establishmentPDOs successrate2023-01NCT06658080 Breast cancerRecruiting Drug testingSensitivityCorrelation2024-11NCT06315868 Breast cancerRecruiting PDOs establishmentSensitivityCorrelation2023-06NCT06459791 Breast cancer/Refractory breast carcinomaRecruiting Drug testingPFS, OS, ORR,2024-01NCT07260188 Invasive breast cancerRecruiting PDOs establishmentPDOs success rate2025-09NCT06155305 Breast cancerRecruiting Drug testingSensitivityCorrelation2023-12NCT05767931 Invasive breast cancerUnknown Drug testingAntitumor rate2023-03NCT05720676Breast cancerUnknownDrug testingSensitivity2023-03NCT03655015 Lung neoplasm Recruiting PDOsestablishmentSensitivity2018-10NCT07239544 Lung cancerActive not recruitingT-cell therapyAntitumor effects2025-07NCT07106528 Small cell lung cancerNot yet recruitingDrug testingPFS, OS, ORR2026-08NCT04826913Non-small cell lung cancerUnknownDrug testingSensitivity2026-04NCT06406660 Small cell lung cancerRecruiting Drug testingORR2024-04NCT06406608 Non-small cell lung cancerRecruiting Drug testingORR, PFS2024-04NCT06405230 Lung cancerNot yet recruitingBiomarker predictionPD-L1 correlation2025-06NCT06959173Lung cancerRecruitingDrug testingORR2025-02NCT05669586Lung cancerRecruitingDrug testingPFS, DCR, OS2023-02NCT06686342 Head and neck squamous cell carcinomaRecruiting Drug testingORR2025-03NCT06755762 Head and neck squamous cell carcinomaRecruitingDrug testing PDOsestablishment2024-08NCT06791941 Head and neck tumor Recruiting PDOs establishmentDrug testingPDCOs successrate2024-03NCT06612827 Undifferentiated carcinoma Recruiting PDOs establishmentDrug testingResponse to treatmentPDOs success rate2025-03NCT05832398 Colorectal cancerUnknown Drug testingPFS, OS2023-05NCT06787625 Colorectal cancerRecruiting Gene function studyorganoid viability2024-05NCT06907342 Colorectal cancerRecruiting Drug testingORR, DCR, CBR2025-05NCT06100016 Colorectal cancerRecruiting Drug testingSensitivity, PFS2023-10NCT05883683 Advanced colorectal carcinomaNot yet recruitingDrug testingSensitivity2024-10NCT05955196 Cancer of the colon Recruiting PDOs establishmentDrug testingPDOs successRateSensitivity2023-01NCT07054086 Esophageal cancer Recruiting PDOs establishmentPDOs success rateCorrelation2025-03NCT06196554 Gastric cancerUnknown Drug testingSensitivityCorrelation2023-05NCT06196554 Gastric cancerUnknown Drug testingSensitivityCorrelation2023-05NCT05442138 Early gastric cancerUnknown Drug testingORR2026-08NCT05352165Rectal cancerUnknownDrug testingPCR2023-01NCT06100003Gastric cancerRecruitingDrug testingSensitivity2023-10NCT05351398Advanced gastric cancerUnknownDrug testingORR2026-01NCT07258186 Endometrial cancerRecruiting PDOs establishmentPDOs success rate2025-03NCT06603506 Endometrial cancer Recruiting PDOs establishmentDrug testingPDOs successRateSensitivity2024-11NCT06786780 Cervical cancerEnrolling by invitationDrug testingORR, CR, PR2023-01NCT06195150Clear cell renal cell carcinomaRecruiting PDOs establishmentPDOs success rate2023-11NCT06662071Bladder cancerRecruitingDrug testingRFS2024-10NCT06227065 Bladder cancerNot yet recruitingDrug testingPathological response2024-10NCT06272240 Ovarian cancerRecruiting PDOs establishmentComposition of the TME2024-01NCT06229522Ovarian cancerRecruitingDrug testingPFS2023-11NCT07029100 Glioblastoma Not yet recruitingPDOs establishmentPDOs success rate2025-07NCT06781372 GlioblastomaNot yet recruitingPDOs establishmentCorrelation2025-04NCT05432518GlioblastomaRecruitingDrug testingOS, PFS2023-06NCT05532397GliomaRecruitingDrug testingPFS, OS2023-02NCT06581549MesotheliomaRecruitingDrug testingSensitivity2023-11NCT06959979 Medulloblasto-mRecruiting Drug testingcell survival2024-11NCT06686979 Salivary gland cancersRecruiting Drug testingORR2025-03NCT06064682 Osteosarcoma Recruiting PDOs establishmentDrug testingPDOs successrateSensitivity2024-02NCT06713057 Thyroid cancerNot yet recruitingDrug testingORR, PFS, OS2024-12This table summarizes interventional and observational trials registered on the ClinicalTrials.gov platform that were identified by querying the terms “cancer/tumor” and “organoid.” Trials initiated from 2023 onward were included to reflect the most recent clinical translation of organoid technologies in oncologyORR Objective response rate, CR Complete response, PR Partial response, PFS Progression-free survival, OS Overall survival, DCR Disease control rate, RFS Relapse-free survival, PCR Pathologic complete response, CBR Clinical benefit rate Clinical trials on PDOs in cancer Drug testing Coherence assessment Sensitivity, Consistency PDOs establishment PDOs success rate PDOs success rate PDOs establishment PDOs success rate PDOs establishment PDOs establishment Drug testing PDOs success Rate Correlation Hepatocellular carcinoma/ Colorectal cancer Cholangiocarcinoma/ Hepatocellular carcinoma/ Pancreatic neoplasm PDOs establishment Drug testing PDOs success rate Sensitivity PDOs establishment Drug testing PDOs success rate Cell survival PDOs establishment, Drug testing PDCOs success rate PDCOs success rate This table summarizes interventional and observational trials registered on the ClinicalTrials.gov platform that were identified by querying the terms “cancer/tumor” and “organoid.” Trials initiated from 2023 onward were included to reflect the most recent clinical translation of organoid technologies in oncology ORR Objective response rate, CR Complete response, PR Partial response, PFS Progression-free survival, OS Overall survival, DCR Disease control rate, RFS Relapse-free survival, PCR Pathologic complete response, CBR Clinical benefit rate Targeted therapy and resistance Beyond predicting therapeutic efficacy, PDOs provide a clinically relevant platform for interrogating the emergence, maintenance, and mechanistic basis of therapeutic resistance. By preserving patient-specific genetic backgrounds, lineage states, and aspects of intratumoral heterogeneity, PDOs are particularly well suited for dissecting tumor-intrinsic determinants of resistance in a clinically relevant context. Multiple studies have demonstrated the capacity of PDOs to reveal resistance-associated signaling dependencies. For example, colorectal cancer organoids harboring wild-type or mutant FGFR4 exhibit differential responses to FGFR4 inhibitors, which suppress ERK1/2, AKT, and STAT3 more effectively than standard chemotherapy [108]. In hepatocellular carcinoma, sorafenib-resistant PDOs show enhanced stemness, EMT, and dedifferentiation phenotypes, implicating the mTOR pathway as a resistance driver [109]. Similarly, in pancreatic cancer, chemoradiation induces SOX2 and OCT4 plasticity, and targeting β-catenin reduces stemness and restores sensitivity [110]. PDOs also enable interrogation of intratumoral heterogeneity in drug resistance. Using single-cell transcriptomics combined with hepatobiliary tumor organoids, Zhao et al. demonstrated that resistant subpopulations collaborate via JAK–STAT–driven metabolic adaptation. Specifically, CD44⁺ cells conferred survival advantage and promoted resistant niche formation [111]. Spatial disparities are also evident, in gastric cancer, organoids from metastatic lymph nodes showed broader chemoresistance than those from primary tumors, highlighting metastasis-associated resistance linked to spatial heterogeneity [112]. In ovarian cancer, platinum resistance profiles from PDOs and their matched xenograft models show strong concordance with patient outcomes [113]. Mechanistically, YBX1 upregulation promotes platinum resistance via the m5C–CHD3/homologous recombination repair axis, enhancing DNA damage tolerance and identifying YBX1 as a potential target [114]. In addition to pathway-level analysis, PDOs support functional biomarker discovery and validation of targetable vulnerabilities. In gastric cancer, resistance to fluorouracil–cisplatin is linked to ADAR1-mediated stabilization of SCD1 mRNA through A-to-I RNA editing, promoting lipid reprogramming and β-catenin activation. Inhibition of SCD1 restores chemosensitivity [115]. In CRC, LGR4-driven Wnt signaling activation upregulates the ferroptosis suppressor SLC7A11, while monoclonal antibody blockade of LGR4 re-sensitizes tumors to therapy [116]. Single-cell analyses have further identified metabolic resistance mediators such as STMN1, VEGFA, and NDRG1 [117]. PDOs also support rational design of combination therapies. In pancreatic cancer, differential PDO responses to FOLFIRINOX regimens were used to guide transcriptomics-based stratification and combination optimization [118]. In residual disease models, combining RNA polymerase I or PI3K inhibitors with BCL-2 family inhibitors significantly suppressed regrowth, offering a strategy to overcome resistance in minimal residual disease settings [119, 120]. Collectively, these studies establish PDOs as versatile platforms for linking resistance phenotypes to their molecular context. Through integration of single-cell analysis, functional genomics, and pharmacologic testing, PDOs support mechanistic insight, biomarker validation, and the development of more personalized therapeutic strategies. Multidimensional assay endpoints and advanced analytics for PDOs Biochemical assays, such as ATP-based viability measurements, remain widely used in high-throughput drug screening to assess overall cytotoxicity in PDOs [121, 122]. However, these approaches are inherently limited by their endpoint-based and bulk nature, which obscures temporal dynamics, clonal heterogeneity, and organoid-level variability. Accordingly, PDO analysis is shifting toward non-destructive, multidimensional platforms that preserve spatial and temporal information. Imaging-based approaches are central to this transition. Brightfield and phase-contrast microscopy enable label-free longitudinal monitoring, but provide limited resolution of three-dimensional architecture and molecular organization [123]. By contrast, fluorescence imaging and confocal microscopy offer spatially resolved viability and structural readouts; luciferase-based 3D assays and calcein-AM/propidium iodide staining, for example, distinguish viable from dead cells while retaining tissue context [124, 125]. Advanced imaging platforms further extend these capabilities. Dekkers et al. developed BEHAV3D, a multispectral 3D imaging system that tracks engineered T-cell interactions with PDOs in real time and links cytotoxic activity to mechanism of action [126]. Likewise, automated high-content imaging in standardized 384-well formats supports scalable and reproducible phenotypic screening across PDO cohorts [127]. In parallel, computational image analysis and machine learning are being integrated into PDO workflows to extract clinically informative features from complex datasets. These approaches quantify morphometric and textural heterogeneity at single-organoid resolution, reconstruct dynamic response trajectories, and show strong associations with patient outcomes, including progression-free survival [128]. Platforms such as OrganoID enable automated segmentation and longitudinal tracking of individual organoids with high concordance to manual annotation, while supporting morphodynamic analysis across multiple cancer types [129] Similarly, Spiller et al. developed an imaging-based high-content workflow for label-free drug-response profiling at single-organoid resolution, enabling detection of dynamic phenotypic heterogeneity beyond conventional bulk assay [130]. Together, the evolution of organoid detection technologies will continue advancing towards greater intelligence, integration, and non-destructiveness. Real-time, dynamic, and non-invasive monitoring technologies will enable us to capture the biological activities of organoids with unprecedented spatiotemporal resolution. Organoid-based immune modeling and applications in immunotherapy The TME consists of immune cells, fibroblasts, vascular endothelial cells, ECM, and soluble factors, functioning as a central regulator of immune evasion and treatment response [131]. Traditional tumor organoids, composed solely of epithelial tumor cells, lack immune and stromal components, limiting their utility in immunotherapy research. To model the cellular complexity and dynamic interactions of the TME in vitro, PDO cultures must be explicitly engineered through co-culture with a range of non-tumor cell types, enabling the stepwise reconstruction of TME components. A conceptual milestone in this area was provided by Dijkstra et al., who demonstrated that co-culture of autologous tumor organoids with peripheral blood lymphocytes could enrich tumor-reactive T cells and enable direct assessment of T-cell-mediated killing of matched PDOs, thereby establishing a patient-specific platform for modeling antitumor immunity [132]. Subsequent studies have extended this framework to incorporate adaptive, innate, and stromal components, including cancer-associated fibroblasts (CAFs) [133, 134], dendritic cells (DCs) [135], CAR-T cells [126], natural killer (NK) cells [136], and tumor-associated macrophages (TAMs) [137]. These systems have established PDO-based immune modeling as a tractable platform for dissecting tumor–immune interactions, immunosuppressive microenvironmental programs, and response to immunotherapy. (Fig. 4). Organoid modeling of immune response and immunotherapy. This schematic illustrates how PDOs serve as platforms to model tumor–immune interactions and guide immunotherapy development. (a) Recreating the tumor microenvironment via co-culture with immune and stromal cells like CAFs, DCs, NK cells, TAMs and T cells. (b) Modeling immunosuppressive mechanisms, including dissection of RTK and CCR5-mediated resistance pathways. (c) Immunotherapy screening platforms, enabling assessment of TILs, CAR-T, NK cells, and ICIs. (d) Engineering immune effector cells, such as TCR-T and NK cells, through expansion and functional profiling. This integrative framework supports preclinical immunotherapy evaluation and personalized strategy optimization Integration of immune and stromal components Immune- and stroma-integrated PDO models have shown that non-malignant TME components are not merely accessory elements, but active regulators of tumor progression, therapeutic resistance, and immune escape. In cholangiocarcinoma PDOs, TAMs promote tumor growth and chemoresistance, implicating macrophage-driven immunosuppression in treatment failure [138]. In CRC, co-culture with CD4⁺ T cells induces regulatory T cell (Treg) polarization, activating transcriptional programs linked to poor clinical outcomes [139]. Similarly, in metastatic CRC, PDO–DC co-culture systems have shown that tumor-derived PGE2 and IL-6 drive cDC2 differentiation toward an immunosuppressive DC3-like state with impaired T-cell priming, consistent with findings in matched patient tissue and blood samples [140, 141]. Stromal incorporation further expands the biological relevance of these models. In CRC, PDO–CAF co-cultures recapitulate primary tumor transcriptomes and reveal JAK/STAT signaling–mediated chemoresistance [142], whereas in pancreatic ductal adenocarcinoma, CAFs promote epithelial–mesenchymal transition (EMT) and invasive phenotypes [133]. More complex multicellular models, such as PDO–CAF–myeloid triple co-cultures, have further reconstructed TAM-like immunosuppressive states and phagocytic reprogramming [143]. In parallel, engineering refinements, including microfluidic PDO-on-chip systems, provide tighter control over TME composition and improve model reproducibility, thereby enabling functional assessment of sensitivity to immune checkpoint blockade, including anti–PD-L1 agents such as atezolizumab [52]. Together, these studies indicate that incorporation of immune and stromal elements enables PDOs to capture microenvironment-dependent programs that shape treatment resistance and immunotherapy response, moving these models closer to the biological complexity of patient tumors. Organoid models for immunotherapy evaluation Immunocompetent PDO models provide a versatile platform for evaluating immune cell–mediated cytotoxicity and assessing immune checkpoint blockade efficacy. These models enable the examination of tumor-intrinsic and immune-modulated resistance mechanisms in a physiologically relevant context. For example, co-culture of PDOs with T cells or peripheral blood mononuclear cells (PBMCs) has demonstrated enhanced immune-mediated killing upon immune checkpoint inhibitor (ICI) exposure, with responses mediated via both contact-dependent and paracrine mechanisms [144]. Additionally, epigenetic priming in breast cancer PDOs upregulates antigen presentation machinery, thereby increasing T cell cytotoxicity and supporting ICI combination strategies [145]. Similarly, PDO systems using gel–liquid interface (GLI) cultures allow for effective PDO–PBMC interactions, where PD-1 blockade responses correlate with clinical outcomes and circulating tumor-reactive T-cell phenotypes, thereby enabling real-time immunotherapy monitoring [146]. Beyond immune-mediated cytotoxicity, PDO-based systems have provided key insights into tumor-intrinsic resistance mechanisms. In oral mucosal melanoma, PDOs derived from ICI-resistant patients exhibited heightened receptor tyrosine kinase (RTK) signaling, particularly NGFR upregulation, and responded to combined RTK inhibition and PD-1 blockade [147]. In PDAC, macrophage–PDO co-cultures identified a CCL5–Sp1–AREG–YAP feedback loop that reinforces stemness and gemcitabine resistance via CCR5/AKT/Sp1/CD44 signaling. Disrupting this loop restored drug sensitivity across PDOs, xenografts, and clinical samples [148]. These studies underscore the role of TME interactions in modulating therapeutic resistance, showing that PDO co-culture models can recapitulate both immune and intrinsic resistance mechanisms in response to treatment. Moreover, PDOs are increasingly leveraged to isolate, expand, and evaluate tumor-reactive effector cells. Co-culture with autologous tumor-infiltrating lymphocytes (TILs) enables enrichment of CD8⁺ T cells with personalized reactivity. When coupled with single-cell RNA-seq, this approach supports TCR discovery, functional validation, and next-generation TCR-T therapy development [149]. In melanoma, immune-enriched PDOs preserve native immune elements. Treatment with IL-2 and anti–PD-1 expands cytotoxic TIL populations, with navitoclax (a BCL-2 inhibitor) further enhancing antitumor efficacy [150]. Similar applications have been demonstrated in CRC, where PDO–TIL co-cultures provide a platform for cytotoxicity profiling and personalized treatment strategies [151]. In lung cancer and PDAC, PDO–NK cell co-cultures facilitate functional screening of NK-based strategies, revealing tumor-mediated suppression of NK effector function and offering insights into immune evasion mechanisms [152, 153]. Collectively, PDOs have evolved from passive drug-testing models into dynamic platforms for immune cell discovery, functional profiling, and engineering. These models are poised to play a critical role in personalized immunotherapy development, including TCR-T, adoptive TIL therapy, and NK cell-based immunotherapies. Ongoing advances in autologous immune-cell isolation, integration of multi-omics, and the development of standardized safety and quality-control frameworks will be essential for optimizing these models and facilitating their clinical implementation. Organoid-based modeling of radiotherapy response and resistance Radiotherapy is a mainstay for solid tumors, yet clinical benefit is constrained by tumor-intrinsic genetics, cellular state, and microenvironmental modifiers—most notably hypoxia. PDOs preserve patient-specific responses in a three-dimensional context, providing a functional platform to deconvolute determinants of radiosensitivity, enable response stratification, and nominate radiosensitizing combinations with translational potential [154]. Mechanistically, PDO-based studies have mapped actionable radioresistance programs across key stress-response axes. In cervical cancer, 3D-bioprinted PDOs identified the NSUN6–ALYREF–m5C–NDRG1 pathway as a driver of homologous recombination competence and radio-tolerance, with NSUN6 expression associating with treatment outcome [155]. In CRC, functional PDO screens uncovered multiple resistance nodes spanning endoplasmic reticulum stress and DNA repair (PRKCSH) [156], ferroptosis suppression via GPX4 transcriptional activation (FOXP4) [157], and ATM-dependent DNA damage repair (TTI1) [158], collectively supporting PDOs as tractable systems for radiosensitizer target discovery and companion biomarker development. Beyond defining resistance circuitry, PDOs also provide a direct functional link between molecular background, microenvironmental context, and inter-patient heterogeneity in radiation response. In CRC, BRAFV600E-mutant PDOs exhibited relative tolerance to radiotherapy or chemotherapy alone but increased vulnerability to combined chemoradiation, consistent with mutation-guided therapeutic stratification [159]. A head and neck PDO biobank demonstrated strong concordance between in vitro radiation responses and clinical results, indicating that TP53 loss alone is insufficient, whereas HPV16 E6/E7 profoundly influences radiosensitivity through cell-cycle control [160]. Importantly, PDOs can also quantify microenvironmental modifiers: in nasopharyngeal carcinoma, PDO-derived estimates of radiobiological parameters (including α/β and oxygen enhancement ratio) demonstrated that hypoxia substantially elevates the effective dose requirement, informing individualized fractionation and dose optimization [161]. From a translational perspective, PDOs enable rational screening and prioritization of radiosensitizing regimens while supporting standardized, scalable readouts. Park et al. showed that the short-chain fatty acid butyrate selectively enhanced CRC radiosensitivity via FOXO3A-mediated cell-cycle inhibition while protecting normal intestinal epithelium, illustrating the potential to identify regimens balancing efficacy and tissue protection [162]. Targeting acid ceramidase similarly augmented radiosensitivity via p53-dependent apoptosis [163]. In rectal cancer, integrative pipelines combining PDO transcriptomics, single-cell profiling, and CRISPR–Cas9 screening implicated PI3K/AKT/mTOR activation as a key driver of radioresistance; pathway inhibition re-sensitized tolerant PDOs, nominating a clinically actionable axis for combination therapy [164]. Methodologically, moderate-throughput chemo-radiotherapy workflows using ATP-based viability with IC50, AUC, and growth-rate metrics have been established for head and neck and CRC PDOs, improving assay reproducibility and cross-study comparability [165]. Moreover, differentiated head and neck PDOs maintained without WNT enable accelerated radiation testing, offering a pragmatic route toward time-sensitive preclinical prediction [166]. Overall, PDO platforms are converging on a model-informed radiotherapy paradigm: they integrate genotype, cellular plasticity, and microenvironmental constraints to (i) resolve resistance circuitry, (ii) derive predictive biomarkers for functional stratification, and (iii) prioritize radiosensitizing combinations suitable for clinical translation (Fig. 5). Continued maturation of engineered culture systems, functional genomics, and longitudinal multi-omics is expected to accelerate incorporation of organoid-guided decision-making into precision radiotherapy workflows. Schematic diagram of the PDO-based platform for radiotherapy evaluation and sensitization strategy development. PDOs are established from tumor specimens and exposed to irradiation under normoxic or hypoxic conditions. Mechanistic profiling identifies key determinants of radioresistance, including impaired DNA repair, suppressed ferroptosis, and hypoxia-induced ROS imbalance. Mutation-stratified response assessment links genetic alterations, like BRAF, TP53 and HPV16, to differential radiosensitivity. Finally, radiosensitizer screening enables identification of candidate agents, including butyrate, acetylase inhibitors, and PI3K/AKT/mTOR pathway inhibitors, to support rational design of combination radiotherapy strategies Challenges and future development directions for organoid technology PDOs have emerged as a transformative preclinical model with significant potential to replicate tumor biology and drive advancements in personalized medicine. Despite their promise, the path to clinical translation remains challenging, with numerous scientific and technical barriers to overcome. These challenges arise from inherent biological biases in the establishment of organoids, the simplifications of current systems when modeling the complex human tumor microenvironment, and bottlenecks in reproducibility and standardization that hinder the widespread application of PDOs. Overcoming these obstacles is essential for realizing the full clinical potential of PDO-based platforms in precision oncology. Biological and technical constraints of PDO modeling PDOs have been successfully generated from a wide range of cancers commonly encountered in clinical practice, and their application has now been extended to rarer tumor subtypes, such as neuroendocrine tumors and hematologic malignancies, including follicular lymphoma [167, 168]. However, the broad translational deployment of PDO technology remains constrained by intrinsic biological and technical limitations. Among these, establishment bias is a major but often underappreciated source of distortion. The success of PDO derivation is strongly influenced by tumor type, histological subtype, sample quality, and the relative abundance of viable epithelial tumor cells [169, 170]. In general, epithelial tumors are established more readily than sarcomas or rare tumors [171, 172]. A known limitation in the clinical application of PDOs is establishment bias, which refers to the selective success of organoid formation from certain tumor types. Tumors with low cellularity, high stromal content, or substantial necrosis are less likely to form viable organoids [173]. As a result, PDO cohorts may overrepresent biologically aggressive subtypes while underrepresenting more differentiated or therapy-sensitive populations, thereby affecting the generalizability of drug testing and biomarker studies. In addition, prolonged culture can induce clonal selection and genetic drift, further altering representation of the original lesion. Challenges in translating organoid models for clinical application The value of PDOs is best understood alongside other human ex vivo tumor models. Compared with tumor slice cultures and tissue explants, PDOs are more scalable, more compatible with high-throughput screening, and better suited to longitudinal perturbation studies, including clonal evolution and acquired drug resistance. By contrast, tumor slices preserve native architecture, cellular diversity, and short-term transcriptional activity, making them useful for rapid functional assessment and early immunotherapy evaluation [174, 175]. However, its relatively short culture time limits its application in longitudinal studies. Air–liquid interface (ALI) culture further expands the ex vivo modeling toolkit. In tissue slice systems, ALI conditions can improve short-term tissue preservation and culture success [176]. ALI-based strategies have also been adapted for organoid culture, where they can enhance the establishment efficiency of selected tumor types, including rare malignancies such as malignant giant-cell tumors and epithelioid sarcomas [177, 178]. Together, these considerations support a shift from reliance on a single model toward a multi-model validation framework, in which tumor slices or explants are used for rapid functional assessment, while PDOs are established in parallel for long-term studies of clonal evolution, drug resistance, and large-scale drug screening. More broadly, clinical translation of organoid models remains constrained by three major barriers: insufficient incorporation of extrinsic components such as immune cells, fibroblasts, and vasculature; culture-associated drift and clonal selection; and limited standardization across matrices, media, workflows, and reporting. Addressing these barriers will require system-level engineering rather than incremental model modification. Emerging approaches include organoid-on-chip platforms with controlled stromal and immune integration, early-passage cryopreservation to preserve clonal diversity, and longitudinal quality-control pipelines incorporating imaging and multi-omics. Together, these strategies aim to improve biological fidelity, reproducibility, and interpretability for clinical implementation. Advancements and future directions in organoid engineering Ethical imperatives and the commitment to animal welfare (the 3R principles) have long been a driving force to develop alternative models in biomedical research [179]. Recent regulatory developments have increased interest in PDOs as human-relevant new approach methodologies (NAMs) in oncology drug development [180]. In particular, the FDA Modernization Act 2.0 and subsequent FDA initiatives have encouraged the use of organoids, organ-on-chip systems, and other advanced in vitro methods to supplement, and in selected contexts reduce, reliance on conventional animal studies [181]. A landmark validation of this paradigm shift occurred in October 2025, with the FDA’s approval of an Investigational New Drug (IND) application for a combination therapy of BAL0891 with immune checkpoint inhibitors. Critically, this represents the first global study where pivotal efficacy data supporting the IND were generated exclusively from human vascularized organoid models, without reliance on traditional animal efficacy trials. Despite this progress, the full integration of NAMs like PDOs as primary tools to replace animal experiments faces considerable challenges. The key to transforming PDOs from a powerful academic research tool into a regulatory-grade NAM lies in establishing robust, validated, and standardized workflows. This encompasses standardized procedures for organoid derivation, culture, quality control, and multidimensional efficacy readouts. Recognizing this need, various industry consortia and academic bodies have begun publishing technical standards. For instance, the Chinese Society of Research Hospitals Association has released “Guidelines for the Construction, Quality Control, and Preservation of Human Hepatobiliary Tumor Cell Organoids” (T/CRHA 018-2023) [182]. The widespread adoption and regulatory recognition of such harmonized standards are essential for the establishment of organoids. Clinical deployment depends on scalable, standardized manufacturing with fit-for-purpose quality control (QC). Automation and high-throughput screening (HTS) workflows increasingly improve process control and batch consistency, and platforms such as 384-well–compatible 3D bioprinting and perfusion systems, as well as acoustic hydrogel-based cell assembly, provide feasible routes for scale-up [183, 184]. In parallel, GMP-aligned workflows and tissue-engineering strategies are beginning to formalize manufacturing and release criteria [185]. A central challenge remains balancing biological fidelity with throughput; tiered standardization, in which model complexity and QC thresholds are aligned with intended use, may offer a practical path forward. The value of organoids lies in modeling dynamic interactions between tumor-intrinsic programs, extrinsic constraints, and therapy-induced pressures. However, current studies still focus predominantly on static, single-time-point phenotypes, thereby overlooking transitional states associated with therapeutic resistance and immune escape. In addition, microbiome-associated regulation of tumor behavior remains underrepresented in most current organoid systems. Notably, Pleguezuelos-Manzano et al. used human intestinal organoids to demonstrate that prolonged exposure to genotoxic pks+ E. coli induces a distinct colibactin-associated mutational signature that is also detectable in a subset of human colorectal cancers, thereby providing direct evidence that organoid platforms can resolve microbe-driven mutagenic processes [186]. Longitudinal models integrating perturbation with multi-omics will therefore be important for reconstructing evolutionary trajectories and causal response pathways. In parallel, synthetic biology and advanced bioengineering are expanding organoid design beyond conventional morphogen- and biomaterial-based approaches. Trentesaux et al. highlighted programmable control of cell communication, adhesion, and cell fate as a key emerging direction [187], whereas Mitrofanova et al. provided a representative example through bioengineered human “mini-colons” with improved longevity, in vivo-like spatial organization, and enhanced cellular complexity [188]. Together, these advances are expected to accelerate the transition of organoids from experimental models toward model-informed precision oncology. Conclusions In summary, tumor organoid technology is transitioning from a model innovation to a broadly applicable translational paradigm. The central challenge has moved beyond feasibility toward achieving an optimal balance among biological complexity, engineering controllability, and clinical relevance. Through interpretable TME reconstruction, longitudinal evolutionary modeling, standardized manufacturing and reporting, and deep interdisciplinary integration, organoids are positioned to evolve from laboratory tools into critical pillars of precision cancer therapy. As these platforms mature, they are expected to play increasingly important roles in tumor biology research, drug development, and individualized treatment planning, thereby supporting the advancement of precision oncology. Acknowledgements The figures of manuscript were drawn by Biorender.com. Abbreviations Cancer‑associated Fibroblasts Clinical Benefit Rate Colorectal Cancer Complete Response Decellularized ECM Disease Control Rate Dendritic Cells Extracellular Matrix Epithelial–mesenchymal Transition Gelatin Methacrylate Gel–liquid Interface High-throughput Screening Immune Checkpoint Inhibitor Mesenchymal Stem Cells Investigational New Drug Metastatic CRC cellsNatural Killer Cells Overall Survival Objective Response Rate Pancreatic Ductal Adenocarcinoma Patient-derived Organoids Patient-derived Xenografts Polyethylene Glycol Progression-free Survival Partial Response Progression-free Survival Pathologic Complete Response Quality Control Relapse-free Survival Three-dimensional Two-dimensional Tumor Microenvironment Tumor-infiltrating Lymphocytes Tumor-associated Macrophages Regulatory T cell Authors’ contributions MB, ML and SW wrote the manuscript and prepared all figures; MR, JY, CZ and YW searched for literature; JL and YX collected the data; FD, DL and HL revised the manuscript; DL and HL provided funding support. All authors read and approved the final manuscript. Funding This work was supported by grants from the National Natural Science Foundation of China (82203823), the Natural Science Foundation of Chongqing (CSTB2024NSCQ-QCXMX0051), and Chongqing medical scientific research project (Joint project of Chongqing Health Commission and Science and Technology Bureau) (2024ZDXM008). Data availability No datasets were generated or analyzed during the current study. Declarations Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Mingzhen Bai, Maoyun Liu and Sheng Wang contributed equally to this work. Contributor Information Fengsheng Dai, Email: cqmudfs@163.com. Dewei Li, Email: lidewei406@sina.com. Hui Li, Email: lih@cqu.edu.cn. References Associated Data Data Availability Statement No datasets were generated or analyzed during the current study.