Patient-Derived Tumor Xenograft Models: Toward the Establishment of Precision Cancer Medicine

From bedside to bench: New insights in epilepsy-associated tumors based on recent classification updates and animal models on brain tumor networks

Low-grade neuroepithelial tumors (LGNTs), particularly those with glioneuronal histology, are highly associated with pharmacoresistant epilepsy. Increasing research focused on these neoplastic lesions did not translate into drug discovery; and anticonvulsant or antitumor therapies are not available yet. During the last years, animal modeling has improved, thereby leading to the possibility of generating brain tumors in mice mimicking crucial genetic, molecular and immunohistological features. Among them, intraventricular in utero electroporation (IUE) has been proven to be a valuable tool for the generation of animal models for LGNTs allowing endogenous tumor growth within the mouse brain parenchyma. Epileptogenicity is mostly determined by the slow-growing patterns of these tumors, thus mirroring intrinsic interactions between tumor cells and surrounding neurons is crucial to investigate the mechanisms underlying convulsive activity. In this review, we provide an updated classification of the human LGNT and summarize the most recent data from human and animal models, with a focus on the crosstalk between brain tumors and neuronal function.

Methods to study xenografted human cancer in genetically diverse mice

Xenografting human cancer tissues into mice to test new cures against cancers is critical for understanding and treating the disease. However, only a few inbred strains of mice are used to study cancers, and derivatives of mainly one strain, mostly NOD/ShiLtJ, are used for therapy efficacy studies. As it has been demonstrated when human cancer cell lines or patient-derived tissues (PDX) are xenografted into mice, the neoplastic cells are human but the supporting cells that comprise the tumor (the stroma) are from the mouse. Therefore, results of studies of xenografted tissues are influenced by the host strain. We previously published that when the same neoplastic cells are xenografted into different mouse strains, the pattern of tumor growth, histology of the tumor, number of immune cells infiltrating the tumor, and types of circulating cytokines differ depending on the strain. Therefore, to better comprehend the behavior of cancer in vivo, one must xenograft multiple mouse strains. Here we describe and report a series of methods that we used to reveal the genes and proteins expressed when the same cancer cell line, MDA-MB-231, is xenografted in different hosts. First, using proteomic analysis, we show how to use the same cell line in vivo to reveal the protein changes in the neoplastic cell that help it adapt to its host. Then, we show how different hosts respond molecularly to the same cell line. We also find that using multiple strains can reveal a more suitable host than those traditionally used for a "difficult to xenograft" PDX. In addition, using complex trait genetics, we illustrate a feasible method for uncovering the alleles of the host that support tumor growth. Finally, we demonstrate that Diversity Outbred mice, the epitome of a model of mouse-strain genetic diversity, can be xenografted with human cell lines or PDX using 2-deoxy-D-glucose treatment.

Patient-Derived Xenografts: A Valuable Preclinical Model for Drug Development and Biomarker Discovery

Patient-derived xenografts (PDXs), established by implanting patient tumor cells into immunodeficient mice, offer a platform for faithfully replicating human tumors. They closely mimic the histopathology, genomics, and drug sensitivity of patient tumors. This chapter highlights the versatile applications of PDXs, including studying tumor biology, metastasis, and chemoresistance, as well as their use in biomarker identification, drug screening, and personalized medicine. It also addresses challenges in using PDXs in cancer research, including variations in metastatic potential, lengthy establishment timelines, stromal changes, and limitations in immunocompromised models. Despite these challenges, PDXs remain invaluable tools guiding patient treatment and advancing preclinical drug development.

Moving low-risk myelodysplastic syndromes from humans to mice: is it truly that simple?

Not available.

Models in Pancreatic Neuroendocrine Neoplasms: Current Perspectives and Future Directions

Pancreatic neuroendocrine neoplasms (pNENs) are a heterogeneous group of tumors derived from multiple neuroendocrine origin cell subtypes. Incidence rates for pNENs have steadily risen over the last decade, and outcomes continue to vary widely due to inability to properly screen. These tumors encompass a wide range of functional and non-functional subtypes, with their rarity and slow growth making therapeutic development difficult as most clinically used therapeutics are derived from retrospective analyses. Improved molecular understanding of these cancers has increased our knowledge of the tumor biology for pNENs. Despite these advances in our understanding of pNENs, there remains a dearth of models for further investigation. In this review, we will cover the current field of pNEN models, which include established cell lines, animal models such as mice and zebrafish, and three-dimensional (3D) cell models, and compare their uses in modeling various disease aspects. While no study model is a complete representation of pNEN biology, each has advantages which allow for new scientific understanding of these rare tumors. Future efforts and advancements in technology will continue to create new options in modeling these cancers.

Generation, evolution, interfering factors, applications, and challenges of patient-derived xenograft models in immunodeficient mice

Establishing appropriate preclinical models is essential for cancer research. Evidence suggests that cancer is a highly heterogeneous disease. This follows the growing use of cancer models in cancer research to avoid these differences between xenograft tumor models and patient tumors. In recent years, a patient-derived xenograft (PDX) tumor model has been actively generated and applied, which preserves both cell-cell interactions and the microenvironment of tumors by directly transplanting cancer tissue from tumors into immunodeficient mice. In addition to this, the advent of alternative hosts, such as zebrafish hosts, or in vitro models (organoids and microfluidics), has also facilitated the advancement of cancer research. However, they still have a long way to go before they become reliable models. The development of immunodeficient mice has enabled PDX to become more mature and radiate new vitality. As one of the most reliable and standard preclinical models, the PDX model in immunodeficient mice (PDX-IM) exerts important effects in drug screening, biomarker development, personalized medicine, co-clinical trials, and immunotherapy. Here, we focus on the development procedures and application of PDX-IM in detail, summarize the implications that the evolution of immunodeficient mice has brought to PDX-IM, and cover the key issues in developing PDX-IM in preclinical studies.

Improving the predictive power of xenograft and syngeneic anti-tumour studies using mice humanised for pathways of drug metabolism

Drug development is an expensive and time-consuming process, with only a small fraction of drugs gaining regulatory approval from the often many thousands of candidates identified during target validation. Once a lead compound has been identified and optimised, they are subject to intensive pre-clinical research to determine their pharmacodynamic, pharmacokinetic and toxicological properties, procedures which inevitably involve significant numbers of animals - mainly mice and rats, but also dogs and monkeys in much smaller numbers and for specific types of drug candidates. Many compounds that emerge from this process, having been shown to be safe and efficacious in pre-clinical studies, subsequently fail to replicate this outcome in clinical trials, therefore wasting time, money and, most importantly, animals. Due to high rates of metabolism and a differing spectrum of metabolites (some pharmacologically active) in rodents, species differences in drug metabolism can be a major impediment to drug discovery programmes and confound the extrapolation of animal data to humans. To circumvent this, we have developed a complex transgenic mouse model – 8HUM - which faithfully replicates human Phase I drug metabolism (and its regulation), and which will generate more human-relevant data from fewer animals in a pre-clinical setting and reduce attrition in the clinic. One key area for the pre-clinical application of animals in an oncology setting – almost exclusively mice - is their use in anti-tumour studies. We now further demonstrate the utility of the 8HUM mouse using a murine melanoma cell line as a syngeneic tumour and also present an immunodeficient version 8HUM_Rag2 -/- - for use in xenograft studies. These models will be of significant benefit not only to Pharma for pre-clinical drug development work, but also throughout the drug efficacy, toxicology, pharmacology, and drug metabolism communities, where fewer animals will be needed to generate more human-relevant data.

Data augmentation and multimodal learning for predicting drug response in patient-derived xenografts from gene expressions and histology images

Patient-derived xenografts (PDXs) are an appealing platform for preclinical drug studies. A primary challenge in modeling drug response prediction (DRP) with PDXs and neural networks (NNs) is the limited number of drug response samples. We investigate multimodal neural network (MM-Net) and data augmentation for DRP in PDXs. The MM-Net learns to predict response using drug descriptors, gene expressions (GE), and histology whole-slide images (WSIs). We explore whether combining WSIs with GE improves predictions as compared with models that use GE alone. We propose two data augmentation methods which allow us training multimodal and unimodal NNs without changing architectures with a single larger dataset: 1) combine single-drug and drug-pair treatments by homogenizing drug representations, and 2) augment drug-pairs which doubles the sample size of all drug-pair samples. Unimodal NNs which use GE are compared to assess the contribution of data augmentation. The NN that uses the original and the augmented drug-pair treatments as well as single-drug treatments outperforms NNs that ignore either the augmented drug-pairs or the single-drug treatments. In assessing the multimodal learning based on the MCC metric, MM-Net outperforms all the baselines. Our results show that data augmentation and integration of histology images with GE can improve prediction performance of drug response in PDXs.

The application of zebrafish patient-derived xenograft tumor models in the development of antitumor agents

The cost of antitumor drug development is enormous, yet the clinical outcomes are less than satisfactory. Therefore, it is of great importance to develop effective drug screening methods that enable accurate, rapid, and high-throughput discovery of lead compounds in the process of preclinical antitumor drug research. An effective solution is to use the patient-derived xenograft (PDX) tumor animal models, which are applicable for the elucidation of tumor pathogenesis and the preclinical testing of novel antitumor compounds. As a promising screening model organism, zebrafish has been widely applied in the construction of the PDX tumor model and the discovery of antineoplastic agents. Herein, we systematically survey the recent cutting-edge advances in zebrafish PDX models (zPDX) for studies of pathogenesis mechanisms and drug screening. In addition, the techniques used in the construction of zPDX are summarized. The advantages and limitations of the zPDX are also discussed in detail. Finally, the prospects of zPDX in drug discovery, translational medicine, and clinical precision medicine treatment are well presented.

Immune organoids: from tumor modeling to precision oncology

Cancer immunotherapies, particularly immune checkpoint inhibitors, are rapidly becoming standard-of-care for many cancers. The ascendance of immune checkpoint inhibitor treatment and limitations in the accurate prediction of clinical response thereof have provided significant impetus to develop preclinical models that can guide therapeutic intervention. Traditional organoid culture methods that exclusively grow tumor epithelium as patient-derived organoids are under investigation as a personalized platform for drug discovery and for predicting clinical efficacy of chemotherapies and targeted agents. Recently, the patient-derived tumor organoid platform has evolved to contain more complex stromal and immune compartments needed to assess immunotherapeutic efficacy. We review the different methodologies for developing a more holistic patient-derived tumor organoid platform and for modeling the native immune tumor microenvironment.

Improving the predictive power of xenograft and syngeneic anti-tumour studies using mice humanised for pathways of drug metabolism

Drug development is an expensive and time-consuming process, with only a small fraction of drugs gaining regulatory approval from the often many thousands of candidates identified during target validation. Once a lead compound has been identified and optimised, they are subject to intensive pre-clinical research to determine their pharmacodynamic, pharmacokinetic and toxicological properties, procedures which inevitably involve significant numbers of animals - mainly mice and rats, but also dogs and monkeys in much smaller numbers and for specific types of drug candidates. Many compounds that emerge from this process, having been shown to be safe and efficacious in pre-clinical studies, subsequently fail to replicate this outcome in clinical trials, therefore wasting time, money and, most importantly, animals. The poor predictive power of animal models in pre-clinical studies is predominantly due to lack of efficacy or safety reasons, which in turn can be attributed mainly to the significant species differences in drug metabolism between humans and animals. To circumvent this, we have developed a complex transgenic mouse model – 8HUM - which faithfully replicates human Phase I drug metabolism (and its regulation), and which will generate more human-relevant data [REFINEMENT] from fewer animals [REDUCTION] in a pre-clinical setting and reduce attrition in the clinic. One key area for the pre-clinical application of animals in an oncology setting – almost exclusively mice - is their use in anti-tumour studies. We now further demonstrate the utility of the 8HUM mouse using a murine melanoma cell line as a syngeneic tumour and also present an immunodeficient version 8HUM_Rag2-/- - for use in xenograft studies. These models will be of significant benefit not only to Pharma for pre-clinical drug development work, but also throughout the drug efficacy, toxicology, pharmacology, and drug metabolism communities, where fewer animals will be needed to generate more human-relevant data.

Co-targeting of specific epigenetic regulators in combination with CDC7 potently inhibit melanoma growth

Melanoma is a highly aggressive skin cancer that frequently metastasizes, but current therapies only benefit some patients. Here, we demonstrate that the serine/threonine kinase cell division cycle 7 (CDC7) is overexpressed in melanoma, and patients with higher expression have shorter survival. Transcription factor ELK1 regulates CDC7 expression, and CDC7 inhibition promotes cell cycle arrest, senescence, and apoptosis, leading to inhibition of melanoma tumor growth and metastasis. Our chemical genetics screen with epigenetic inhibitors revealed stronger melanoma tumor growth inhibition when XL413 is combined with the EZH2 inhibitor GSK343 or BRPF1/2/3 inhibitor OF1. Mechanistically, XL413 with GSK343 or OF1 synergistically altered the expression of tumor-suppressive genes, leading to higher apoptosis than the single agent alone. Collectively, these results identify CDC7 as a driver of melanoma tumor growth and metastasis that can be targeted alone or in combination with EZH2 or BRPF1/2/3 inhibitors.

Patient-derived xenograft (PDX) models, applications and challenges in cancer research

The establishing of the first cancer models created a new perspective on the identification and evaluation of new anti-cancer therapies in preclinical studies. Patient-derived xenograft models are created by tumor tissue engraftment. These models accurately represent the biology and heterogeneity of different cancers and recapitulate tumor microenvironment. These features have made it a reliable model along with the development of humanized models. Therefore, they are used in many studies, such as the development of anti-cancer drugs, co-clinical trials, personalized medicine, immunotherapy, and PDX biobanks. This review summarizes patient-derived xenograft models development procedures, drug development applications in various cancers, challenges and limitations.

Identification of Prognostic Markers of Gynecologic Cancers Utilizing Patient-Derived Xenograft Mouse Models

Patient-derived xenografts (PDXs) are important in vivo models for the development of precision medicine. However, challenges exist regarding genetic alterations and relapse after primary treatment. Thus, PDX models are required as a new approach for preclinical and clinical studies. We established PDX models of gynecologic cancers and analyzed their clinical information. We subcutaneously transplanted 207 tumor tissues from patients with gynecologic cancer into nude mice from 2014 to 2019. The successful engraftment rate of ovarian, cervical, and uterine cancer was 47%, 64%, and 56%, respectively. The subsequent passages (P2 and P3) showed higher success and faster growth rates than the first passage (P1). Using gynecologic cancer PDX models, the tumor grade is a common clinical factor affecting PDX establishment. We found that the PDX success rate correlated with the patient’s prognosis, and also that ovarian cancer patients with a poor prognosis had a faster PDX growth rate (p < 0.0001). Next, the gene sets associated with inflammation and immune responses were shown in high-ranking successful PDX engraftment through gene set enrichment analysis and RNA sequencing. Up-regulated genes in successful engraftment were found to correlate with ovarian clear cell cancer patient outcomes via Gene Expression Omnibus dataset analysis.

An overview of kinase downregulators and recent advances in discovery approaches

Since the clinical approval of imatinib, the discovery of protein kinase downregulators entered a prosperous age. However, challenges still exist in the discovery of kinase downregulator drugs, such as the high failure rate during development, side effects, and drug-resistance problems. With the progress made through multidisciplinary efforts, an increasing number of new approaches have been applied to solve the above problems during the discovery process of kinase downregulators. In terms of in vitro and in vivo drug evaluation, progress was also made in cellular and animal model platforms for better and more clinically relevant drug assessment. Here, we review the advances in drug design strategies, drug property evaluation technologies, and efficacy evaluation models and technologies. Finally, we discuss the challenges and perspectives in the development of kinase downregulator drugs.

Using Spheroids as Building Blocks Towards 3D Bioprinting of Tumor Microenvironment

Cancer still ranks as a leading cause of mortality worldwide. Although considerable efforts have been dedicated to anticancer therapeutics, progress is still slow, partially due to the absence of robust prediction models. Multicellular tumor spheroids, as a major three-dimensional (3D) culture model exhibiting features of avascular tumors, gained great popularity in pathophysiological studies and high throughput drug screening. However, limited control over cellular and structural organization is still the key challenge in achieving in vivo like tissue microenvironment. 3D bioprinting has made great strides toward tissue/organ mimicry, due to its outstanding spatial control through combining both cells and materials, scalability, and reproducibility. Prospectively, harnessing the power from both 3D bioprinting and multicellular spheroids would likely generate more faithful tumor models and advance our understanding on the mechanism of tumor progression. In this review, the emerging concept on using spheroids as a building block in 3D bioprinting for tumor modeling is illustrated. We begin by describing the context of the tumor microenvironment, followed by an introduction of various methodologies for tumor spheroid formation, with their specific merits and drawbacks. Thereafter, we present an overview of existing 3D printed tumor models using spheroids as a focus. We provide a compilation of the contemporary literature sources and summarize the overall advancements in technology and possibilities of using spheroids as building blocks in 3D printed tissue modeling, with a particular emphasis on tumor models. Future outlooks about the wonderous advancements of integrated 3D spheroidal printing conclude this review.

Cell Models for Chromosome 20q11.21 Amplification and Drug Sensitivities in Colorectal Cancer

Background and objectives: The chromosome locus 20q11.21 is a commonly amplified locus in colorectal cancer, with a prevalence of 8% to 9%. Several candidate cancer-associated genes are transcribed from the locus. The therapeutic implications of the amplification in colorectal cancer remain unclear. Materials and Methods: Preclinical cell line models of colorectal cancer included in the Cancer Cell Line Encyclopedia (CCLE) collection were examined for the presence of amplifications in 20q11.21 genes. Correlations of the presence of 20q11.21 amplifications with gene essentialities and drug sensitivities were surveyed on salient databases for determination of therapeutic leads. Results: A significant subset of colorectal cancer cell lines in the CCLE (12 of 63 cell lines, 19%) bear amplifications of genes located at 20q11.21. Cancer-associated genes of the locus include ASXL1, DNMT3B, BCL2L1, TPX2, KIF3B and POFUT1. These genes are all amplified in the 12 cell lines, but they are variably over-expressed at the mRNA level, compared to non-amplified lines. 20q11.21 amplified cell lines are sensitive to various tyrosine kinase inhibitors and are resistant to chemotherapy drugs targeting the mitotic apparatus and microtubules. CRISPR and RNAi dependencies screening revealed, besides the β-catenin and KRAS genes, a few recurrent gene dependencies in more than one cell line, including YAP1 and JUP. Conclusions: Cell line models of colorectal cancer with 20q11.21 gene amplifications display dependencies on the presence of specific genes and resistance or sensitivity to specific drugs and drug categories. Observations from in vitro models may form the basis for clinical drug development in this subtype of colorectal cancer. Genetic lesions conferring synthetic lethality to certain drugs or categories of drugs could be discovered with this approach.

Zebrafish Patient-Derived Xenografts Identify Chemo-Response in Pancreatic Ductal Adenocarcinoma Patients

Simple Summary

Treating the PDAC (pancreatic ductal adenocarcinoma) zPDXs (zebrafish patient-derived xenografts) with chemotherapy regimens commonly used, we performed a co-clinical trial testing the predictiveness of the model. We found that zPDX may predict patient outcomes, classifying them into responders (R) and non-responders (NR), reporting a statistically significant higher cancer recurrence rate at 1 year after surgery in the NR group: 66.7 versus 14.3%. Our zPDX model seems to be a promising tool for the stratification of PDAC patients. This is a crucial starting point for future study involving more patients to obtain a method to really personalize the oncological treatment of PDAC patients.

Abstract

It is increasingly evident the necessity of new predictive tools for the treatment of pancreatic ductal adenocarcinoma in a personalized manner. We present a co-clinical trial testing the predictiveness of zPDX (zebrafish patient-derived xenograft) for assessing if patients could benefit from a therapeutic strategy (ClinicalTrials.gov: XenoZ, NCT03668418). zPDX are generated xenografting tumor tissues in zebrafish embryos. zPDX were exposed to chemotherapy regimens commonly used. We considered a zPDX a responder (R) when a decrease ≥50% in the relative tumor area was reported; otherwise, we considered them a non-responder (NR). Patients were classified as Responder if their own zPDX was classified as an R for the chemotherapy scheme she/he received an adjuvant treatment; otherwise, we considered them a Non-Responder. We compared the cancer recurrence rate at 1 year after surgery and the disease-free survival (DFS) of patients of both groups. We reported a statistically significant higher recurrence rate in the Non-Responder group: 66.7% vs. 14.3% (p = 0.036), anticipating relapse/no relapse within 1 year after surgery in 12/16 patients. The mean DFS was longer in the R-group than the NR-group, even if not statistically significant: 19.2 months vs. 12.7 months, (p = 0.123). The proposed strategy could potentially improve preclinical evaluation of treatment modalities and may enable prospective therapeutic selection in everyday clinical practice.

Generation of Non-Small Cell Lung Cancer Patient-Derived Xenografts to Study Intratumor Heterogeneity

Simple Summary

It is widely thought that tumors are composed of different subpopulations of cancer cells carrying genetic alterations with some of them being common among all cells while others are unique for each subpopulation. This variable genetic profile of tumor cells is a component of what is collectively described as intratumor heterogeneity (ITH). Surviving the immune system and therapies, and establishing metastases are forces of natural selection that act upon ITH and drive tumor evolution and, eventually, the clinical presentation of patients. The aim of this prospective study was to investigate ITH in early-stage operable non-small cell lung cancer. We directly grafted human tumors in immunosuppressed mice and compared the genetic profile of the tumors grown in mice with that of the original human tumors. We identified clinical factors that affected the ability of human tumors to grow as mouse xenografts.

Abstract

Recent advances in sequencing technologies have allowed the in-depth molecular study of tumors, even at the single cell level. Sequencing efforts have uncovered a previously unappreciated heterogeneity among tumor cells, which has been postulated to be the driving force of tumor evolution and to facilitate recurrence, metastasis, and drug resistance. In the current study, focused on early-stage operable non-small cell lung cancer, we used tumor growth in patient-derived xenograft (PDX) models in mice as a fast-forward tumor evolution process to investigate the molecular characteristics of tumor cells that grow in mice, as well as the parameters that affect the grafting efficiency. We found that squamous cell carcinomas grafted significantly more efficiently compared with adenocarcinomas. Advanced stage, patient age and primary tumor size were positively correlated with grafting. Additionally, we isolated and characterized circulating tumor cells (CTC) from patients’ peripheral blood and found that the presence of CTCs expressing epithelial-to-mesenchymal (EMT) markers correlated with the grafting potential. Interestingly, exome sequencing of the PDX tumor identified genetic alterations in DNA repair and genome integrity genes that were under-represented in the human primary counterpart. In conclusion, through the generation of a PDX biobank of NSCLC, we identified the clinical and molecular properties of tumors that affected growth in mice.

Large-scale pharmacogenomic studies and drug response prediction for personalized cancer medicine

The response rate of most anti-cancer drugs is limited because of the high heterogeneity of cancer and the complex mechanism of drug action. Personalized treatment that stratifies patients into subgroups using molecular biomarkers is promising to improve clinical benefit. With the accumulation of preclinical models and advances in computational approaches of drug response prediction, pharmacogenomics has made great success over the last 20 years and is increasingly used in the clinical practice of personalized cancer medicine. In this article, we first summarize FDA-approved pharmacogenomic biomarkers and large-scale pharmacogenomic studies of preclinical cancer models such as patient-derived cell lines, organoids, and xenografts. Furthermore, we comprehensively review the recent developments of computational methods in drug response prediction, covering network, machine learning, and deep learning technologies and strategies to evaluate immunotherapy response. In the end, we discuss challenges and propose possible solutions for further improvement.

Multi-parametric characterization of drug effects on cells

We present here a novel multi-parametric approach for the characterization of multiple cellular features, using images acquired by high-throughput and high-definition light microscopy. We specifically used this approach for deep and unbiased analysis of the effects of a drug library on five cultured cell lines. The presented method enables the acquisition and analysis of millions of images, of treated and control cells, followed by an automated identification of drugs inducing strong responses, evaluating the median effect concentrations and those cellular properties that are most highly affected by the drug. The tools described here provide standardized quantification of multiple attributes for systems level dissection of complex functions in normal and diseased cells, using multiple perturbations. Such analysis of cells, derived from pathological samples, may help in the diagnosis and follow-up of treatment in patients.

New technologies in developing recombinant-attenuated bacteria for cancer therapy

Cancer has always been a global problem, with more cases of cancer patients being diagnosed every year. Conventional cancer treatments, including radiotherapy, chemotherapy, and surgery, are still unable to bypass their obvious limitations, and developing effective targeted therapies is still required. More than one century ago, the doctor William B. Coley discovered that cancer patients had tumor regression by injection of Streptococcus bacteria. The studies of cancer therapy using bacterial microorganisms are now very widespread. In particular, the facultative anaerobic bacteria Salmonella typhimurium is widely investigated as it can selectively colonize different types of tumors, locally deliver various antitumor drugs, and inhibit tumor growth. The exciting antitumor efficacy and safety observed in animal tumor models prompted the well-known attenuated Salmonella bacterial strain VNP20009 to be tested in human clinical trials in the early 21st century. Regrettably, no patients showed significant therapeutic effects and even bacterial colonization in tumor tissue was undetectable in most patients. Salmonella bacteria are still considered as a promising agent or vehicle for cancer therapy. Recent efforts have been focused on the generation of attenuated bacterial strains with higher targeting for tumor tissue, and optimization of the delivery of therapeutic antitumor cargoes into the tumor microenvironment. This review will summarize new technologies or approaches that may improve bacteria-mediated cancer therapy.