Emerging organoid models: leaping forward in cancer research

Predominantly, tumor invasion is regarded as a single-cell process. However, recent discoveries have implied that tumor invasion behaves as a cohesive multicellular unit, which is referred to as collective invasion [44]. Cancer organoids have been used as an optimizing model system to reveal the underlying mechanisms of collective invasion. For example, breast cancer organoids were used to investigate the role of leader cells that guide tumor cell invasion and intravasation. By using a live-cell microscopy assay, researchers found that BC organoids with the invasive phenotype extended multicellular strands of cancer cells into the extracellular matrix when the collective invasion was initiated by the specialized cancer cells that expressed K14 and p63 [45]. Similarly, by using cancer organoids, the researchers revealed that the cathepsin B led to the collective invasion in salivary adenoid cystic carcinoma [46], the inhibition of rho-associated protein kinase 2 (ROCK2) associated with initiating collective invasion in colorectal adenocarcinomas [47], and the loss of heat-shock factor 2 (HSF2) correlated with collective invasion in prostate cancer [48]. Moreover, extracellular matrix (ECM) in the tumor microenvironment, such as collagen I, could also modulate collective invasion in colon cancer organoids [49]. These studies exemplify that cancer organoid provides a trackable and reliable means to investigate tumor invasion.

Cancer organoids have also been used to identify the critical mutations that contribute to metastasis formation. Researchers have developed gene-edited CRC organoids carrying only the tumorigenesis driver pathway mutations APC, SMAD4, TP53, KRAS, and/or PIK3CA. These CRC organoids merely formed micrometastases when implanted into the spleen of mice. In contrast, the organoids with both chromosomal instability (CIN) and the tumorigenesis driver pathway mutations were capable of forming large metastatic tumors when transplanted into the mice [38]. These results suggested that CIN played an important role in modulating tumor cells to acquire metastatic behaviors in the CRC. In addition, cancer organoids could also aid in the discovery of critical targets for inhibiting tumor metastasis. In one study, Chandhoke, A.S. et al. discovered that the sumoylation of the PIAS3-Smurf2 pathway could inhibit the invasiveness of mammary tumor organoids [50]. Thus, the organoids provide an effective cancer model to study the mechanisms in promotion and inhibition of tumor invasion.

Anoikis refers to apoptosis of cancer cells induced by insufficient cancer-matrix interactions [51]. Anoikis resistance may allow the survival and proliferation of cancer cells and may contribute to tumor invasion and metastasis. Recently, intestinal organoids were used to study the effect of the RHOA mutation on the dissociation-induced apoptosis. Wang K. et al. genetically edited intestinal organoids with the RHOA mutations, which existed in approximately 14.3% of diffuse-type gastric cancer patients. Then, these organoids were dissociated into single cells. As expected, the RHOAmutation could lead to a higher efficiency of organoids recovery. More importantly, organoids carrying the RHOA mutation showed a better survival time and proliferative capacity, while the wild-type organoids were dead completely when without addition of the inhibitor of anoikis [25]. This result implied that the RHOA mutation could help cancer organoids escape from anoikis.

Metastatic dormancy is a leading cause of cancer recurrence [52]. However, the mechanisms of tumor metastatic dormancy and reactivation are still poorly understood. Cancer organoids have been demonstrated as a useful tool for tumor dormancy studies. Hattar, R. et al. demonstrated that tamoxifen could modulate cancer dormancy in a BC organoid model by reducing the fibronectin level in the extracellular matrix (ECM). BC organoids cultured on the tamoxifen-treated ECM displayed a smaller and smoother morphology compared to the BC organoids cultured on the tamoxifen-untreated ECM. Furthermore, they also found that tumor cell motility and invasion were suppressed by the tamoxifen treatment. These results were consistent with the previous clinical finding that increasing fibronectin level was associated with the lower survival rate in BC patients [53, 54]. Similarly, the antibodies to human collagen I can modulate the tumor dormancy in the BC organoid model by reducing the activity of collagen I in the ECM [55]. These results indicated that the ECM components in the tumor microenvironment could regulate tumor dormancy. In brief, cancer organoids can be used as a tool enabling effective screening of drug candidates that potentially prevent tumor recurrence.

Recent studies have demonstrated that PDTOs can capture the cancer-specific genetic alterations, gene expression, and histopathology in individual patients, which makes them suitable for personalized drug screening [9, 10, 30]. Sachs N and his colleagues constructed BC organoids from surgically resected specimens from 155 cancer patients. By comparing the therapeutic responses of anticarcinogen in the BC organoids and the corresponding patients, they found that the sensitivity to tamoxifen in the BC organoids was closely correlated with that in the original patients with metastatic BC [10]. More recently, personalized hepatocellular carcinoma organoids derived from needle biopsies were used to optimize drug dose for eight patients. The PDTOs displayed a distinct dose-dependent response to the sorafenib treatment in the different patients, which implied the potential value of PDTO models to predict patient-specific drug sensitivities to the targeted drugs [11]. Additionally, cancer organoids also act as an effective tool for interrogating gene-drug association. For example, Saito Y and colleagues constructed cancer organoids from surgically resected specimens from the patients with biliary tract carcinoma. They found that the TP53 mutant organoids were not sensitive to nutlin-3a, while the wild-type organoids were highly sensitive to nutlin-3a [12]. Similarly, the CRC organoids with the TP53 mutation was found insensitive to nutlin-3a [9]. These results agreed well with the clinical outcome in cancer patients with TP53 mutation.

Though the adoptive cell transfer and immunomodulatory checkpoint blockade have shown clear clinical benefit in the long-lasting anti-tumor immune responses, a large proportion of patients is insensitive to immunotherapy due to the heterogeneity of T cell repertoire and human leukocyte antigen (HLA) resulted from patient-specific neo-antigens [58‐60]. Recent advances in tumor organoids offer a promising approach to generate tumor-reactive T cells. For example, Dijkstra KK et al. performed a coculture of tumor organoids with the patient’s peripheral blood lymphocytes. Under the stimulation of tumor organoids, tumor-reactive T cells with patient-specific immunogenic mutations were enriched and expanded, and then they could recognize and kill the autologous tumor organoids [13]. In addition, Finnberg NK et al. demonstrated that cancer organoids culturing at the air-liquid interface (ALI) could directly maintain the native tumor microenvironment for up to 44 days [14]. Furthermore, Neal JT and his colleagues indicated that the established tumor organoids using the ALI method could recapitulate the intrinsic tumor T-cell receptor spectrum and anti-PD-1/PD-L1-dependent human tumor-infiltrating lymphocyte (TIL) activation [15]. Meanwhile, cancer organoids have been used to study the effectiveness of combination immune therapy. Della Corte CM et al. investigated the efficacy of combining the anti-PD-L1 antibody with MEK inhibitor (MEK-I) or the anti-PD-1/PD-L1 therapy alone in non-small cell lung cancer (NSCLC) organoids. The research suggested that the combination therapies had a significantly higher drug response rate than the monotherapy owing to the increase of cell toxicity and immunoreactivity by the induction effect of MEK-I [16]. Notably, there are two clinical trials registered on the website of ClinicalTrials.​gov, involving cancer organoids for immunotherapy (ClinicalTrials.​gov number NCT03778814, NCT02718235).

Overall, these results indicate that cancer organoid culture is a promising system to generate tumor-reactive T cells, to predict immunotherapy sensitivity, and to examine combination immunotherapy.

Cancer organoids have been utilized as a platform to discover cancer prognosis-related hallmarks. Broutier L et al. discovered 30 potential tumor biomarkers by systematically comparing transcriptional differences between healthy organoid lines and primary liver cancer (PLC) organoid lines. Among these 30 tumor biomarkers, 19 genes were associated with PLC in clinical, and within 13 genes were related to poor prognosis in clinical. The researchers further analyzed the remaining 11 genes using The Cancer Genome Atlas (TCGA) and identified three genes associated with poor prognosis in hepatocellular carcinoma and one gene associated with poor prognosis in cholangiocarcinoma. Interestingly, STMN1 overexpression, which was previously thought to be associated with poor prognosis in only hepatocellular carcinoma, was proven here to be associated with low survival in cholangiocarcinoma in clinical [28]. These studies exemplify the potential value of PDTOs for tumor prognostic biomarker discovery.

The PDTOs provide a promising approach for personalized anti-cancer therapy in clinical. According to the studies registered on the website of the ClinicalTrials.​gov as of November 1, 2019, there were 30 projects (1 terminated and 29 ongoing projects) related to cancer organoids. Among these trials, 53% were the observational studies and 47% belonged to the interventional studies, including one trial in phase I and five trials in phase II. Meanwhile, we noted that 73% projects aimed at studying anti-cancer therapy, including tailoring treatments for patients, identifying effective drug combinations, examining T-cell immunotherapy, and evaluating radiotherapy sensitivity; 13% projects aimed to generate patient-derived cancer organoid models; and the remaining projects focused on the mechanistic investigation of cancer onset and progression. Notably, these clinical trials involved a wide range of cancer types, including lung, pancreatic, prostatic, breast, esophagogastric, hepatocellular, biliary tract, neuroendocrine, and colorectal cancers, astrocytoma, and sarcoma [61].

In one clinical trial in the UK, Vlachogiannis G et al. carried out a phase I/II clinical trials to evaluate the clinical value of PDTOs in personalized anti-cancer therapy. In this trial, 71 patients with CRC or gastroesophageal cancer were recruited. Cancer organoids derived from patients’ biopsies displayed the 100% sensitivity, 93% specificity, 88% positive predictive value, and 100% negative predictive value, compared to the drug responses in the corresponding patients [62]. This study provided an encouraging proof that PDTOs can be employed as a clinically relevant model for anti-cancer therapy. Overall, we expect that the PDTO will revolutionize the conventional paradigm of anti-cancer therapy from systemic to individual approaches.

A notable strategy is to generate organoid-on-a-chip by combining organoid with organ-on-a-chip. Organ-on-a-chip is a microfabricated device with integrated living cells, ECM, and microstructures to emulate partial aspects of organ or tissue in their cytoarchitecture, cellular population, and functions [63]. Organ-on-a-chip is featured with the capacities for precise microenvironment control, continuous flow perfusion culture, and high-throughput format. Notably, organ-on-a-chip allows integration of multiple mini-organs in the different microchambers interconnected via microfluidic channels to form human microphysiological system, which provides a unique platform to study cancer multiorgan metastasis via the circulatory system. Nevertheless, at present, most of the organ-on-a-chip systems utilize primary cell lines or stem-cell-derived cells as the cell source to construct organ mimics, and they cannot emulate histological and cellular complexity of native organs and tumors [64]. By incorporating multiple organoids into organ-on-a-chip, organoid-on-a-chip can inherit the benefits from both organoid and organ-on-a-chip and provide an effective tool to study tumor multiorgan metastases and cancer-microenvironment interactions.

A 3D vascularized tumor model was constructed on a chip to study the mechanism of multiorgan metastasis from breast cancer (Fig. 3a). In this chip, endothelial cells (ECs), mesenchymal stem cells (MSCs), and osteoblast-differentiated cells (OBs) were cultured in 3D ECM to mimic bone marrow and muscle microenvironments with the microvascular networks. Extravasation rates of these metastatic BC cells were investigated on these microenvironments with or without adenosine treatment. The result showed that metastatic BC cells displayed distinct extravasation rates in different microenvironments, and blockage of A₃ adenosine receptor in BC cells resulted in increased extravasation rate in the muscle microenvironment [65]. In another study, a four-organ-on-a-chip system was developed to model metastasis of primary lung cancer to the downstream organs, including the brain, liver, and bone (Fig. 3b) [66]. The results implied the metastasis displayed spatiotemporal heterogeneity over the different organs and ultimately led to the damages on all these four organs. However, these tumor-on-chip models were built with cancer cell lines and could not represent the critical features of the native tumor. In turn, incorporation of metastatic tumor organoids with other normal organoids on a chip presents a better way for studying cancer multiorgan metastasis.

Another strategy is to develop sophisticated organoid culture systems for multiple tumor types by using 3D bioprinting. 3D bioprinting allows precise control over spatial heterogeneity in the tumor microenvironment by spatially deterministic deposition of predefined biobanks that may contain multiple cell types, biochemical factors, and ECM (Fig. 3c) [67‐70]. For example, Grolman, J.M., et al. constructed a BC microenvironment to study the role of paracrine signaling network in the regulation of breast cancer metastasis. Breast adenocarcinoma (MDA-MB-231) and macrophages (RAW 264.7) were printed in the hydrogels with distinct spatial distributions and variable geometrical shapes by extrusion-based 3D bioprinting technique (Fig. 3d) [71]. The results indicated that geometric cues regulated the paracrine loop between BC cells and macrophages, which further initiated BC tumor intravasation into the bloodstream. Another example of an in vitro cervical tumor model was established to demonstrate the epithelial-to-mesenchymal transition (EMT). The HeLa cells were mixed with hydrogel and further be fabricated into cell-biomaterial constructs with grid shape by employing an extrusion-based 3D bioprinter (Fig. 3e) [72]. The results implied the supplement of TGF-β-induced EMT and this promoting effect was inhibited by the treatment of disulfiram and EMT pathway inhibitor C19 in a dose-dependent manner, which suggested that the tumor metastasis in 3D culture was a comprehensive result involving the complex interacions between tumor cells, ECM, and 3D microenvironment.