Novel organoid model in drug screening: Past, present, and future

Abstract

Advances in the field of stem cells have led to the development of a technology called organoids. Organoids are cell cluster structures formed by the cultivation of stem cells in a three-dimensional environment in vitro, and they can simulate the living environment of cells in vivo. Organoids play an important role in the screening of drugs for tumor therapy. Compared with traditional drug screening models, tumor organoid models derived from patient tumors have higher sensitivity, heterogeneity, and stability and can restore the real situation of tumors more effectively. Researchers have conducted a number of researches on the feasibility of using organoid technology in drug screening. By testing and comparing the effects of antitumor drugs in organoids and primary tumors, we can select the most appropriate treatment drugs for patients. In the past ten years, organoids from dozens of tissues and biological sample banks from several main organs have been established, and a large number of anticancer drugs have been screened out. This article summarizes the advantages and disadvantages of traditional drug screening models, discusses the development history of organoid technology, and reviews the research results on organoids from tumor drug screening. In addition, the combination of organoid technology and other modern biotechnologies is put forward to further promote the role of organoid technology in the medical field. Finally, this article reviews the history, progress, and prospect on organoids from the view of antitumor drug screening.

1. Introduction

In recent years, although numerous advances have been achieved in the field of tumor treatment, tumors remain an important factor that endangers human life and health. The key to tumor treatment lies in the early detection and the use of effective drugs. In terms of drugs used in treatment, the existence of heterogeneity and the complexity of cancer genome render the treatment effect unideal. How to find the most suitable drug for patients has become the key to tumor treatment, and thus, accurate and individualized drug screening has also gained significance. Before starting the treatment, an accurate and individualized drug screening can be used to select the most effective treatment options with minimal adverse reactions based on the patient symptoms, thereby significantly improving the survival rate of tumor patients; notably, its realization relies heavily on an accurate drug sensitivity screening and testing system.¹ Cell models are the most used detection platforms; however, they cannot simulate various endogenous structural and physiological changes, and this difference often leads to a variety of uncertain factors in clinical trials that use selected drugs. Animal models cannot also be used for the accurate completion of drug sensitivity test because of the loss of tumor heterogeneity during culture, low transplant success rate, long test cycles, etc.² Therefore, numerous standby drugs that have significant effects on cell and animal models have not been applied to clinical settings, resulting in the wastage of considerable research resources. In addition, patients receive ineffective treatment.³ However, the advent of organoid technology has provided a reliable platform for drug screening.

In 2014, Lancaster and Knoblich⁴ systematically proposed the concept of organoids, providing a reliable theoretical basis for the development of biomedicine and disease awareness. Organoids can be derived from normal and tumor tissues. Therefore, when applied to drug development, drugs that specifically target tumor cells can be screened without damaging normal cells. The high fidelity of organoids, stable genomes of multiple passages, and short culture cycles render them with a huge advantage in drug development and clinical guidance.

2. Traditional cancer model in drug screening

At present, the commonly used models for tumor drug screening include patient-derived cancer cell line (PDC) and patient-derived xenograft (PDX) models.

2.1. PDC model

In 1951, as a result of the establishment of the HeLa cell line, the PDC model became a primary model in tumor biology research. Given its simple culture conditions and infinite proliferation in vitro, the PDC model is suitable for large-scale drug screening. The Cancer Cell Line Encyclopedia provides a reference for large-scale drug screening.⁵ This database contains a collection of gene expression information, chromosome copy numbers, and gene sequencing information from 1457 tumor cell lines. However, the PDC model also has its drawbacks. The lack of diversified cell types, spatial organization, and microenvironment similar to the body will cause adverse effects on the culture of stem cells.⁶ The characteristics of tumor cells in the body and their heterogeneity are also lost during the culture in vitro. The prostate tumor cell line is an example.⁷ Several common genetic mutations are absent in the cell lines cultured in vitro, including speckle-type BTB/POZ protein (SPOP) mutation, chromodomain helicase DNA-binding protein 1 (CHD1) deletion, forkhead box protein A1 (FOXA1) mutation, etc. Additionally, tumor cell lines cultured in vitro have poor responsiveness to drugs and thus may prevent newly developed drugs in clinical trials from achieving the same results as in vitro and produce unknown adverse reactions.^8,9

2.2. PDX model

The PDX model plays a very important role in the clinical treatment of tumor patients. This model can rapidly expand clinical samples in vitro and identify drug targets for different tumors through a variety of omics tests and biological information analysis.¹⁰ The PDX model is also an important tool for detecting new antibodies, especially in immune checkpoint blockade therapy. Zhang et al.¹¹ successfully established a non-small-cell lung cancer PDX model by using human peripheral blood cells, which are present in the human immune system, and it was used to evaluate the efficacy of new programmed cell death 1 (PD-1)/programmed cell death ligand 1 (PD-L1) antibodies. At present, the most important issue in the development of oncology drugs is the low success rate of new drug development, mainly given the impossible accurate prediction of the effect of newly developed drugs on patients. The PDX model has a higher predictive value and can better simulate the therapeutic effect of drugs on tumors. Large-scale PDX biobanks are also gradually being established. The established breast cancer PDX biobanks are mainly used for high-throughput drug screening in vitro.¹² The Public Repository of Xenografts (PRoXe) is a public biobank of PDXs for leukemia and lymphoma, and it integrates different tumor types, periods of tumor transcriptomics, and proteomics biomarkers. The PRoXe biobank is being used to conduct clinical studies on various anticancer drugs.¹³

With the development of tumor research, the complexity of tumors exposes the disadvantages of PDX models gradually. Vaccination, passaging, drug screening, and other processes last for long periods (4–8 months); thus, PDX models are unsuitable for patients with extremely advanced or highly invasive tumors.¹⁴ Although the PDX model can retain the tumor heterogeneity to a large extent, the growth of certain cell subpopulations during the passage process will lead to the heterogeneity of original tumors in several patients.¹⁵ In addition, during the tumor culture process, the matrix of the animal model used will replace the human matrix component gradually,¹⁶ and in turn, the human tumor microenvironment will not be able to reproduce well. The interaction between human tumor tissue and the rapidly infiltrating animal matrix microenvironment will affect the paracrine signals of tumors. Furthermore, the PDX model has defects, such as large demands of tumor samples, low transplant success rate, high cost, and lack of a functional immune system.

3. Novel organoid model in drug screening

3.1. Development of organoids

In the past decade, numerous achievements have been made in the field of stem cells, among which the rapid development of organoid systems has become an important technology in the medical field and presents considerable application prospects. Organoids refer to the culture of tissue stem cells in vitro; they maintain the stable function of original stem cells, allowing them to continue dividing and differentiating, and finally form micro-organisms similar to the source organ in terms of genes, structure, spatial distribution, function, etc.¹⁷ Organoids are an organic combination of different types and functions of cells; they simulate well the spatial position, function, and growth status of organs in the body. Rheinwald and Green¹⁸ first described human cell culture methods in 1975. In 1977, they first used artificially cultivated human stem cells to successfully complete three-dimensional (3D) tissue reorganization.¹⁹ In 2009, Sato et al.²⁰ cultured crypt cells isolated from mouse intestinal segments in a 3D Matrigel to form a crypt-villi complex, which was similar to the microstructure of the intestine. In 2013, organoid technology was selected by Science as one of the top 10 breakthroughs in technological development.²¹ In 2015, human brain organoids were hailed as one of the top 10 breakthrough technologies by MIT Technology Review. In 2017, organoid technology was again named an annual technology in the field of life sciences by Nature Methods. At present, organoids are mainly derived from pluripotent stem cells (PSCs) and organ-specific adult stem cells.^22,23 Stem cells are induced and differentiated into organoids with different structures and functions through the induction of various growth factors. In the past decade, more than 4000 research results on organoids have been attained. More than a dozen organoids, such as those of the colon, brain, lung, stomach, esophagus, prostate, pancreas, liver, breast, and bladder, exist, and among them, the number of intestinal organoids is the largest, accounting for nearly 30% (data from Web of Science).

3.2. Research progress on tumor organoids in drug screening

Patient-derived organoids (PDOs) can be obtained by the 3D culture of a patient's tumor tissue in vitro. PDOs highly summarize the characteristics of the tumor source, retain the heterogeneity between individuals, and have a high application value in medicine. PDOs can be used in functional tests, such as high-throughput drug screening. Pauli et al.²⁴ established 56 PDOs from the tumor tissue of 769 patients, with a model success rate of 39%, thus enabling a high-throughput drug screening.

3.2.1. Intestinal PDOs

In 2011, Sato et al.²⁵ successfully cultured intestinal adenomas, metaplastic Barret epithelium, and colon cancer tissues for the first time in vitro and established tumor-derived organs derived from patients. Later, researchers studied these tumors. Extensive research has been performed on organoid stability, heterogeneity, and drug sensitivity. Van de Wetering and colleagues²⁶ established the first living organoid biobank, and these researchers have conducted an experiment in which 22 strains of colon cancer organoids from different patients were tested using 83 experimental drugs; the result showed that tumor organoids of different genetic backgrounds have various sensitivities to drugs. In addition, according to their research, all the subsequencing analyses proved that the colon tumor organoid model retains the heterogeneity of the original tumor, and that tumor organoids can maintain the genome stability during long-term culture. Their research showed that organoids can highly restore the heterogeneity of colon tumors and different sensitivities to drugs, whereas tumor organoids play an important role in drug screening for different patients. Weeber et al.²⁷ developed a new tumor organoid model culture condition, and gene sequencing was performed on 1977 cancer-related genes of 14 colon tumor organoids and their original tumors. They observed that 90% of the two gene mutations were preserved. Experiments also showed that organoids function highly to reduce the heterogeneity of the original tumor and can be stabilized. Verissimo et al.²⁸ used the genetic diversity of tumor organoid banks to study the RAS gene in colorectal cancer in vitro and observed that the RAS gene mutation has a strong internal relationship with certain tumor drugs, providing a basis for drug screening. Emmink et al.²⁹ revealed that ATP-binding cassette sub-family B member 1 (ABCB1)^positive/aldehyde dehydrogenase (ALDH)^low tumor cells can enable the pumping of irinotecan (topoisomerase 1 inhibitor) into the organoid cavity to protect ABCB1^positive/ALDH^low tumor-initiating cells from being killed by drugs. Fiore et al.³⁰ discovered that compared with primary tumor cells, rimonabant can significantly alleviate the toxic and side effects of oxaliplatin and 5-fluorouracil in organoid models and reduce DNA fragmentation. Rimonabant can also retain the same gene between organoids and primary tumor cells, which indicates that tumor organs are highly sensitive in screening drugs and can maintain their genetic stability. In 2018, 21 pairs of colorectal tumor organoid/primary tumor drug sensitivity data were included in the literature published in Science.³¹ Researchers calculated the statistical results of predicting clinical drugs by tumor organoids: 100% sensitivity, 93% specificity, 88% positive prediction (that is, 88% of the drugs that work on tumor organoids also affect primary tumors), and 100% negative prediction (i.e., all drugs that show no effect on tumor organoids are ineffective on primary tumors). This study proved that the drug sensitivity between tumor organoids and primary tumors is highly consistent. The above studies have also demonstrated the effectiveness of the application of tumor organoids in intestinal tumors. A series of advantages, such as high fidelity, genome stability, and short culture time, make tumor organoids stand out in drug screening for intestinal tumors.

3.2.2. Liver tumor organoids

In general, three types of cell sources including induced PSCs (iPSCs), cancer cell lines, and primary cells, have been used to culture liver organoids to identify their cellular and molecular mechanisms and to validate and test drugs. Compared with iPSCs and cancer cell lines, primary hepatocyte cells are most similar to in vivo liver in disease modeling and drug screening.³² These cells can more accurately represent donor tissues with 10-fold fewer base substitutions in their genes than iPSCs.³³ Importantly, they also preserve the donor-specific histological architecture (such as pseudo-glands), gene expression patterns (such as alpha-fetoprotein in expression profiles of tumor), and genetic alterations, making them ideal for patient- and disease-specific studies.³⁴

Hepatic organoids were first established in 2013.³⁵ In 2017, based on the classic liver organoid culture protocol, Broutier et al.³⁴ removed R-spondin-1, Noggin, and Wnt3a, added dexamethasone and Rho kinase inhibitors, and increased the digestion time of liver tumor tissues in the experiment. They successfully cultured hepatocellular and bile duct tumor organoids for the first time. When culturing liver tumor organoids, Broutier et al.³⁴ also observed that primary liver tumors with different gene mutation patterns exhibited varying sensitivities to therapeutic drugs. Catenin beta 1 mutant hepatocyte-derived tumor organoids are resistant to the Wnt inhibitor LGK974, but bile duct-derived tumor organoids are sensitive to it. This finding is consistent with the drug sensitivity of tumors in patients, thus indicating that tumor organoids can mimic the sensitivity of actual tumors in patients. In addition, during a drug screening of tumor organoids, sorafenib, the only approved drug for first-line treatment of hepatocellular carcinoma, showed therapeutic prospects for certain types of cholangiocarcinoma.³⁶

In drug screening, whether drug toxicity can be accurately judged often plays an important role in whether a drug can be used in clinics. Drug-induced liver toxicity is generally mediated by cytochrome P450 enzymes (CYP450s), and liver organoids can induce CYP450s to normal physiological levels in inducing differentiation.³⁷ Therefore, it is necessary to use liver organoids to test drug hepatotoxicity in preclinical trials. Similarly, iPSC-derived cardiac organoids are used to test the cardiotoxicity of drugs,³⁸ whereas iPSC-derived renal organoids are used to detect renal toxicity.³⁹ The CYP450s in novel organoids grown on perfusable chips showed a considerably higher activity than those in the organoids grown in static conditions.⁴⁰ In this acetaminophen-induced toxicity study,⁴⁰ the organoid-on-a-chip exhibited sensitivity and applicability for drug testing. Such system may provide a cost-efficient and simple platform for drug screening.

3.2.3. Prostate and pancreatic tumor organoids

Pancreatic organoids were first reported in 2013.⁴¹ In 2015, Boj et al.⁴² successfully cultured pancreatic tumor organoids. The establishment of this tumor cell presented difficulty due to the large stromal component of the tumor. Huang et al.⁴³ used gemcitabine and other epigenetic inhibitors to treat pancreatic tumor organoids from five patients. The results showed that the five samples had different sensitivities to the drug, and the pancreatic organoids can retain the sensitivity of tissues from different patients to the drugs in vitro. Although only five samples were used in this study, it proved the feasibility of using pancreatic tumor organoids in the treatment of pancreatic tumors. Prostate organoids were first reported in 2014,⁴⁴ the same year when Gao et al.⁷ established prostate tumor organoids. Enzalutamide and abiraterone acetate are new-generation prostate cancer treatment drugs, and they have become the standard treatment for advanced prostate cancer.⁴⁵ Both drugs can significantly prolong the survival time of patients with prostate cancer^46,47; however, their therapeutic effectiveness is very unstable, and about 30% of patients can sustain the treatment effect for more than 6 months.⁴⁸ Therefore, a model that can accurately simulate a patient's sensitivity to therapeutic drugs need to be developed. Gao et al.⁷ established seven types of advanced prostate organoids in vitro to test their effects on the sensitivity to enzalutamide and abiraterone acetate and observed that their therapeutic effect coincided with that in vivo. In the above test, a tumor organoid was derived from circulating tumor cells (CTCs) of patients with prostate cancer. Therefore, a patient-specific organoid model can be established with a small amount of blood from patients without them having to undergo complex and invasive tissue biopsy.

3.2.4. Gastric and esophageal tumor organoids

In 2014, gastric organoids were successfully cultured for the first time,⁴⁹ followed by the successful culture of gastric tumor organoids in the following year.⁵⁰ In 2019, Seidlitz et al.⁵¹ reported for the first time the successful establishment of a biological sample bank of gastric tumor organoids, which contained 20 different patient samples. In this gastric tumor organoid biobank, the tumor organoids present two states: a hollow cystoid glandular structure and a solid spherical structure with diffusely distributed cells, consistent with the Lauren classification of gastric cancer. In the same case of gastric cancer, the tissue structure of tumor organoids was consistent with that of primary tumors, and the expression levels of gastric cancer markers, such as carcinoembryonic antigen (CEA), cytokeratin 7 (CK7), and periodic acid-Schiff (PAS), were consistent. The researchers observed that the tumor organoids amplified by ERBB2 gene responded well to the monoclonal antibody, trastuzumab. Meanwhile, tumor organoids with c-KIT gene deletion are sensitive to imatinib, and tumor organoids with cyclin-dependent kinase inhibitor 2A gene deletion are resistant significantly to palbociclib. These results show that tumor organoids can be used for targeted therapy under the guidance of targets. This biological sample bank has proven that tumor organoids can maintain the characteristics of primary tumor morphology and genomics. In 2018, as reported in Cell, Nanki et al.⁵² successfully established 67 tumor organoids, which by far constitute the largest biological sample bank of gastric cancer organoids. This sample bank studied the epidermal growth factor receptor pathway in detail. This study also has a reference value in targeted medicine for tumor therapy. In addition, the tumor organoids amplified by the MET gene are sensitive to crizotinib, the MET inhibitor, suggesting that corresponding patients can be treated with this drug. The above two biological sample banks show that gastric cancer organoids can be used to guide specific drug treatment of tumors and provide a reliable model for drug screening. In addition, Vlachogiannis et al.⁵³ used biopsy organoid samples to perform drug tests and compared them with the tumor of patients in clinical trials. These researchers discovered that tumor organoids have extremely high sensitivity and specificity and proposed their certain ability to predict efficacy. Li et al.⁵⁴ successfully cultivated esophageal tumor organoids for the first time in 2018. In this trial, the researchers evaluated the advantages of esophageal cancer organoids in 10 patients for drug screening and tested the organoids against 24 patients to determine their sensitivity to anticancer drugs; the results were consistent with clinic in vivo. This study confirmed that using esophageal organoids in medium-throughput drug sensitivity test is feasible. However, no large samples of esophageal cancer organoid biobanks are currently available. The research suggests that esophageal tumor organoids still have shortcomings, such as low success rates, and have certain differences in gene expression between specific subtypes of tumors and tumor organoids. Therefore, large samples and more comprehensive and in-depth studies are urgently needed to improve the application and value of organoid biobanks in the field of esophageal cancer.

3.2.5. Other types of tumor organoids

The successful establishment of breast cancer organoids and bladder cancer organoids was first reported in 2018.^55,56 In two trials, gene mutations and copy number variation of chromosome were consistent between the breast and bladder cancer organoids and primary tumors. Kenny et al.⁵⁷ developed the quantitative high-throughput drug screening system based on the organoid culture technology for the detection of chemotherapy drugs in ovarian cancer. Thus far, organoids derived from various primary tumors have been screened for drugs. All of them played a vital role, providing a reliable basis for personalized treatment of different patients.

4. Summary and future perspectives

Compared with the traditional PDC and PDX models, the PDO model has incomparable advantages. This model reflects well on the specificity of tumors in each patient, simulates the complexity of primary tumors in vitro, and improves the efficiency and accuracy of screening the most suitable drugs for different patients. Given the advantages of the PDO model for the precise treatment of tumors, we combined our own clinical experience and pieces of literature to propose research plans and testing procedures (Fig. 1).

Fig. 1.

Organoids in personalized medicine. First, CTC, biopsy, surgical resection, and relapse specimens from tumor patients were used to construct organoid models. Then, the constructed organoid models and human specimens were sequenced to confirm the consistency of the constructed models. Finally, the treatment schemes of tumor patients, including chemotherapy drugs, radiotherapy effects, multi-point targeted drugs (sorafenib, lenvatinib, etc.), autologous cytokine-induced killer (CIK), autologous natural killer (NK) cells, autologous TILs, and immune checkpoint treatment schemes (PD-1, PD-L1, and CTLA-4), were verified in different PDO models. Based on the results of high-throughput drug screening, we can optimize individualized treatment plans and build a PDO database, thereby providing proper treatment plans for patients with similar genetic results. Abbreviations: CTC, circulating tumor cell; CTLA-4, cytotoxic T-lymphocyte-associated protein 4; PD-1, programmed cell death 1; PD-L1, programmed cell death ligand 1; PDO, patient-derived organoid; TILs, tumor-infiltrating lymphocytes.

Organoid models as tools to simulate other chronic diseases is a promising but unexplored field. The evolution from precancerous lesions to malignancy is a continuous process. Extensive and progressive hepatic histological alterations are encompassed in the process from fatty liver to non-alcoholic steatohepatitis (a critical risk factor for hepatocellular carcinoma), whereas little is known about the underlying mechanisms. Ouchi et al.⁵⁸ developed a reproducible method to derive multi-cellular human liver organoids and recapitulated several key features of steatosis and steatohepatitis, including steatosis, inflammation, and fibrosis phenotypes, in a successive manner under a free fatty acid treatment. However, whether such hepatocyte organoids can reappear the long-term step from hepatic steatosis and steatohepatitis to liver cirrhosis and hepatocellular carcinoma remains to be investigated.

Recently, the problems of organoid technology are gradually being solved. In the past 10 years, organoid technology has achieved growth from single cells to millimeter-level tissues, but it still faces huge challenges compared with solid organs. The main restrictions to the continued growth of organoids are the use of Matrigel and vascularization. Rambani et al.⁵⁹ observed that 0.2 mm cell clusters often cannot obtain sufficient oxygen and nutrition through diffusion. At present, researchers are attempting to create a large substance exchange space by changing the matrix composition. Specifically, researchers are using a sponge-like scaffold instead of Matrigel because the former can provide a larger cavity. Robertson et al.⁶⁰ used a detergent to perfuse and decellularize liver tissue directly and used the vascular network inside the liver to undertake organoid nutrient transport. On the one hand, vascularization of organoids is also a major problem in organoid cultures. Insufficient vascular endothelial growth factor results in a low density of organoid capillaries. Pham et al.⁶¹ reported that functional vascular networks have not been implemented in organoids. The key is to find factors that determine the induction of angiogenesis, especially whether limiting shortcomings, such as the pressure on cells in different organoids, cell pH, blood components, etc., exist inside the cell.

To date, 3D printing technology can be used to print the vascular network of heart tissue at the microscopic level. Meanwhile, microfluidic systems and bioreactors can accurately control the input of oxygen and nutrients from organoids and the timely removal of metabolic wastes. The emergence of small-micelle-mediated human organ efficient clearing and labeling (SHANEL) technology allowed researchers to optimize the accuracy of organ features at the cellular level in 3D images.⁶² This new technology can quickly achieve tissue clearing. SHANEL technology has been used to draw 3D maps of pancreas, kidney, heart, and other organs. If we combine SHANEL and organoid technologies to obtain transparent organoids from patients, we can quickly identify drug targets, determine micro-changes at the cellular level during drug screening, and accurately assess whether anticancer drugs can be used in clinical settings. With the development of modern biotechnology, the problems of organoids are being solved step by step. Tumor organoids will further accurately simulate the situation of primary tumors among patients, thereby improving the accuracy of drug screening. In the future, intelligent and biomimetic materials, microfluidics, and 3D printing technology will be highly integrated with the organoid technology to improve the accuracy, sensitivity, and durability of tumor organoid responses to drugs in vitro, achieve accurate and individualized drug screening, and remarkably improve the cure rate of patients with cancers.

Authors’ contributions

X. Nie and Z. Liang contributed equally to this study. Study concept and design: L. Ye and Y. Yang. Drafting the manuscript: L. Ye, X. Nie, Z. Liang, K. li, H. Yu, and Y. Huang. Obtained funding: L. Ye and Y. Yang. All authors read and approved the final manuscript.

Declaration of competing interest

The authors declare that they have no conflict of interest.