Breast cancer (BC) is a prevalent malignancy among women globally. In 2020, BC surpassed lung cancer to become the most frequently diagnosed malignancy (1). Rapid advances in drug testing and high-throughput sequencing technology in the past few decades have resulted in considerable progress in the early-stage screening and clinical treatment of BC (2). Approximately 70%–80% of early-stage BCs can be cured with surgery and targeted therapy, with the mortality rate of patients with BC decreasing each year. However, approximately 685,000 patients still succumb to late-stage BC each year; therefore, it remains the leading reason for mortality among women worldwide (3).

At present, two-dimensional (2D) tumor cell lines and patient-derived tumor xenografts (PDTXs) remain the commonly used cancer research models (4). Although these experimental models have played a substantial role in basic BC research in the past, limitations remain.

BC emergence and progression involve processes other than tumor cell proliferation; they are intricately associated with the tumor microenvironment (TME) (5, 6), including tumor-associated fibroblasts (7), tumor-associated macrophages (8), adipocytes (9, 10), immune cells (11, 12) and the extracellular matrix (ECM) (13) (Figure 1). However, 2D-cultured tumor cells lack not only the TME but also the genetic heterogeneity of the original tumor owing to the substantial genetic changes that occur in multiple clones (14). Furthermore, although PDTXs can preserve the TME and the complex tumor structure, they also have many limitations, including higher experimental costs, longer experimental time, increased technical requirements, low success rate, and ethical issues (4). Therefore, alternative models for BC research are warranted.

In 2009, Sato et al. illustrated for the first time that a solitary leucine-rich repetitive G-protein-coupled receptor 5 (LGR5)-positive intestinal stem cell can form three-dimensional (3D) epithelial structures. These structures exhibit distinct polarized epithelial characteristics, featuring proliferative crevices and differentiated villous sections, and can maintain stable growth in vitro (15). Subsequently, this culture technique has facilitated the formation of other kinds of epithelial organoids, and patient-derived organoid (PDO) models have been developed from different organs and cancer types (4), including the liver (16), breast (17), colon (18), lungs (19), and stomach (20). Recently, 3D culture technology has emerged as a promising in vitro model for cancer. Different from monolayer cells and PDTXs, PDOs are 3D cultures that are formed in vitro from surgically resected or biopsy tissues, which are structurally and functionally similar to in vivo organs. They not only provide a good summary of the gene expression profiles, personalized characteristics, and cell types of native tissues but also preserve the TME of a patient’s tumor. In addition, organoids possess self-organization and self-renewal properties (14, 21, 22).

In 1979, Emerman et al. cultured normal mammary epithelial cells in collagen gel and observed that different from the 2D culture model, the collagen gel matrix allowed the entry of nutrients to the basolateral cell surface, cells to approach the medium surface and gas phase, the interaction of epithelial cells with matrix elements, and sublaminar flexibility, allowing changes in cell shape. This approach provided a unique microenvironment for breast epithelial cells in a 3D environment (23). In 1982, Bissell et al. studied BC organoids for the first time and noted that the presence of the ECM in 3D models can better simulate gene expression in BC cells (24). This study not only demonstrated the importance of the TME but also shifted the understanding of BC cells from 2D to 3D. Furthermore, in 1982, M Hiratsuka et al. isolated and cultured breast organoids with duct-like structures from breast biopsy tissues (25). In 1990, a study revealed the varying effects of distinct ECM compositions on murine mammary tissue-derived cells (26). In 1992, Bissell reported for the first time that normal primary mammary epithelial cells and biopsied cancer cells could be cultured to generalize the growth behavior and structural and functional differentiation of these cells in vivo (27). In 2007, Bissell et al. revealed that vital cellular signals disappeared when cells were cultured outside their natural environment on flat 2D plastic surfaces. However, a significant portion of these essential microenvironmental cues were reinstated using 3D laminin-rich extracellular matrix cultures (28). Since Sato et al. first cultured intestinal organoids from a single intestinal epithelial stem cell in 2009, BC stem cells have also gradually gained importance as the key to organoid formation (17). In 2012, breast Lgr5+ cells not only served as breast stem cells and survived for a long time in the passage process but also developed normal ductal structures similar to the breast via cytokine arrangement (29). Sachs et al. cultured 95 BC organoids from 155 surgical BC tissues and successfully established the pioneering BC organoid repository; they observed that the histopathology and hormone receptor status of BC organoids were similar to those of the original tumor; furthermore, they noted that BC organoids can be used for in vitro drug screening and respond similarly as in vivo xenografts and patients (19). Recently, with the continuous improvement and optimization of experimental techniques, the success rate of BC organoid culture has reached 87.5% (30). Therefore, BC organoid technology is widely employed in basic research, drug screening, and individualized therapy (Figure 2).

PDO models exhibit various captivating advantages, which transcend those offered by traditional cell and patient-derived xenograft (PDX) models (4, 14, 19).

Tumor heterogeneity refers to the concept that tumor cells can exhibit distinct genetic and phenotypic characteristics, including gene mutation, cell morphology, and gene, protein, and marker expression. Furthermore, because of genetic and phenotypic differences, tumor cells exhibit differences in proliferation, transfer ability, and drug resistance. Tumor heterogeneity can be categorized into two types: inter- and intratumor heterogeneity. Intertumor heterogeneity refers to the diversity observed across distinct tissues. For example, tumor occurrence and development are different in each patient; furthermore, the treatment process is different, which leads to tumor heterogeneity among different patients. On the other hand, intratumor heterogeneity primarily depends on the TME and intrinsic characteristics of tumor cells (35). Compared with monolayer cultures, PDOs can better summarize the heterogeneity of original tumor tissues. Diermeier et al. performed RNA sequencing (RNA-seq) to compare the tumor and organoid transcriptome profiles of luminal B (MMTV-PyMT) and HER2/nue-amplified (MMTV-Neu-NDL) mouse models. They observed that, in most instances, the organoids could effectively replicate the expression profile of the initial tumor tissue. Only a small fraction, approximately 1% of the gene transcriptome (142 of 13,854 expressed genes), exhibited statistically significant differences between the tumor and day 6 organoids. The affected genes were predominantly protein-coding genes (85%), affecting their connected signaling and metabolic pathways. Simultaneously, noncoding RNA levels were largely unaltered (36). Sachs et al. successfully created 95 organoid biobanks from patients’ tumors, representing all BC subtypes. Then, they compared the whole genome sequences of organoids and primary BC tissues via whole genome sequencing and RNA-seq and observed that the transcriptomic, histological, and genomic fingerprints of the original tumor tissue were largely unchanged in the organoids after 20 passages. Furthermore, they performed H&E staining to histopathologically analyze tumor tissues and organoid sections and observed that the phenotype of BC organoids was often similar to that of the original BC tissue. For example, solid organoids are produced for ductal carcinoma, whereas loose organoids are produced for lobular carcinoma. In addition, nuclear atypia is preserved in organoids, including enlarged and polymorphic nuclei and high mitotic activity (19). In the context of the BC model, preserving the key and common specific biomarkers, including estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor-2 (HER2), is vital. These biomarkers serve as vital targets for drugs against BC (37). Chen et al. successfully established organoids from 132 patients with BC and evaluated the status of ER, PR, HER2, and Ki-67 on organoids and tumors via immunohistochemistry. They observed that the expression patterns of these BC markers were well preserved, regardless of whether the organoids had received systemic anticancer therapy (38). Bhatia et al. generated an organoid bank from 83 BC tissues. They not only elucidated that the organoids accurately replicated the traits of the patient’s tumor but also observed that when triple-negative breast cancer (TNBC) organoids were implanted into NOD/SCID mice, the resultant tumors were morphologically similar to the original tumor (39). In a recent study, Saeki et al. developed PDOs from 10 patients with BC and observed that these BC PDOs had different growth rates and patterns. Subsequently, they subjected these PDOs to transcriptome analysis at the single-cell level, innovating novel analytical techniques: (1) clustering cells within each PDO, (2) discerning genes specifically expressed in each cluster, and (3) quantifying their resemblance to subsequently categorize the clusters. These analyses revealed that each PDO comprises multiple subpopulations with different cell states and cycles, estrogenic responses, and epithelial–mesenchymal transition-like gene expression programs. This suggests that PDO-specific cell clusters reflect different PDO properties. Moreover, they subjected two other BC tissues or PDO pairs to RNA-seq and revealed that although a complete one-to-one correspondence was not achieved, PDOs retained some degree of tumor heterogeneity of the original tissue after successive passages (40).

In this review, we primarily discussed the research and application prospects of BC PDO models. PDOs can preserve a part of the TME of the original tissue, cell morphology, tumor heterogeneity, and the patient’s responses to drugs. Moreover, even after several passages, most of their features are preserved. However, the current PDO models remain immature and imperfect. One limitation of the PDO model is that it cannot generalize the entire TME of the patient and lacks a tumor matrix, blood vessels, and immune cells. In response, Neal et al. co-cultured PDOs with other cellular components to reconstruct the TME (60). Furthermore, Davaadelger et al. added fibroblasts to PDOs to investigate the effect of BRCA1 mutations on progesterone in breast cells (72). In addition, extracellular elements and microorganisms may be co-cultured with PDOs in the future (73, 74).

Organoids can be established from patient-derived tumor tissues and can be efficiently expanded and passaged for further drug screening. This makes PDO a valuable model for translating into practical applications and tailoring individualized cancer treatments. One of the most promising applications of PDOs is their ability to predict a patient’s drug response, possibly guiding individualized cancer treatment. For example, Chen et al. successfully guided the individualized treatment of six patients with BC via the drug sensitivity experiments of their PDOs; all patients achieved good efficacy (38). However, PDOs cannot predict the drug responses of the patient’s whole body.

In contrast, organoids include microfluidic systems with advanced 3D bioprinting technologies as the new technology. By employing various bioprinting techniques (such as extrusion, inkjet, photocuring, etc.), the use of bioinks, such as natural or synthetic polymers, cells, or biomolecules, creates complex and anatomically accurate structures and makes each organoid more uniform in volume, which can simulate the biophysical properties of tumors more realistically (75, 76). For example, Heinrich et al. used 3D printing technology to build mini-brain models of glioblastoma and macrophages to study cell interactions and tested the therapies targeting this interaction (77). In addition, the fluid of the microfluidic device can flow and connect to organoids of different tissue origins, further simulating the exchange of nutrients and substances and the formation of vascular networks between tumors. Therefore, the behavior of tumor metastasis and spread can be better simulated (78, 79). Skardal et al. established a “metastasis-on-a-chip” system and simulated the liver metastasis of colon cancer (80). In BC, Swaminathan et al. used bioprinting to directly print 3D spheres of breast epithelial cells and co-culture them with vascular endothelial cells, thereby successfully creating tissue models that could be used for drug efficacy studies (81). However, the accuracy of 3D printing, the cell survival rate, and how to accurately replicate the heterogeneity of the tumor in the model requires further research and validation.

Finally, organoids exhibit considerable potential in regenerative medicine. In addition to being derived from a patient’s tumor tissues, PDOs can also be developed from a patient’s stem cells. Dai et al. noted that breast organoids can be established from human-induced pluripotent stem cells and can differentiate into breast components in immunodeficient mice to further regenerate into breast-like structures. In the future, the issue of BC cell repair and regeneration may be solved after resection (82).

In general, as a new clinical model, PDOs are not only important to study BC mechanisms but also to screen drugs and guide treatment regimens. In the future, research directions can focus on updating breast organoid culture techniques and how to simulate the TME more comprehensively. In conclusion, PDO technology will continue to be an important tool in BC research and clinical drug screening.

YS: Writing – original draft. ZG: Writing – review & editing. GC: Writing – original draft. YN: Writing – original draft. CZ: Writing – review & editing. WL: Writing – review & editing. JL: Writing – review & editing.