Cancer is a leading cause of morbidity and mortality worldwide. In 2015, it was responsible for 8.8 million deaths, making it the second leading cause of death globally (1). In the United States, nearly 40% of people will be diagnosed with cancer within their lifetimes, and the national expenditure for cancer patient care is estimated to reach $156 billion by the year 2020.

A great deal of progress has been made in understanding cancer development over the last century enabling a steady decline in cancer mortality over the past two decades (2). However, in general, there still remains a large gap in survival rates between localized and metastatic disease (3), indicating that advanced disease is more poorly understood and more difficult to treat. In fact, early detection still remains one of the most reliable ways to ascertain an effective treatment response (4). The high mortality of advanced disease calls for a better understanding of cancer cell biology and, in particular, the mediating factors leading to cancer progression. One such realm of cancer biology that is increasingly recognized as a key factor in tumor progression, metastasis, and chemotherapeutic drug resistance is the tumor microenvironment (TME) (5).

Tumors develop in complex and dynamic environments that aid their sustained growth, invasion, and metastasis (6). This space, called the TME, directly surrounds the tumor and consists of an array of non-malignant cells including fibroblasts, immune cells, adipose tissue, blood and lymphatic vessel networks, and signaling molecules of the extracellular matrix. Current research has demonstrated the vital role of the TME in maintaining and progressing the tumor phenotype and as a possible focus for targeted therapy (6). Complex interactions between the TME and malignant cells occur through a very sophisticated network of cellular communication. Many of these signaling pathways operate through direct cell-to-cell contact or by classical paracrine signaling loops of cytokines or growth factors with their receptors. However, more recently, extracellular vesicle (EV) shedding has emerged as another important mechanism of cellular interchange (7). EVs are lipid bilayer-bound vehicles that are released from the cell membrane and carry nucleic acids (DNA, mRNA, and miRNA), proteins, and lipids to neighboring or distant cells (8). EVs exhibit wide-ranging roles in maintaining normal cellular and biological physiology. Undoubtedly, however, the most heavily researched area of EV-associated pathology is their role in tumor development and chemotherapeutic resistance in cancer. In this review, we highlight the most current studies involving EVs and their influence on the pathogenesis of cancer.

Cells reside in diverse microenvironments that help maintain physiological order within a tissue. During tumorigenesis, this region consists of tumor cells and an array of non-malignant cells including fibroblasts, immune cells, adipose tissue, blood and lymphatic vessel networks, and the extracellular matrix called the TME (5). In fact, non-cancerous cells can account for greater than 50% of the overall tumor mass (9). Within the TME, complex communication pathways are occurring among its inhabitants (10), allowing the tumor and the TME to independently influence one another and co-evolve. It is apparent that the TME has a vital role in maintaining and progressing the tumor phenotype and that it is a possible focus for targeted therapy (6).

A healthy local microenvironment surrounding a tumor may initially help to provide anticancer effects. This was first demonstrated in melanoma where the number of tumor-infiltrating lymphocytes was shown to correlate with better clinical outcomes (11). This concept has also been demonstrated in many other malignancies including colorectal cancer, where an active TME and signs of an immune response were associated with the absence of early metastasis and increased survival (12). In fact, the notion that local immune cells act as guards against nascent transformed cells was developed over 60 years ago (13).

However, as tumors develop, their microenvironment becomes disrupted, protection is lost, and tumor progression may be allowed to continue. The association between a disrupted microenvironment and malignancy is well established and is illustrated during the process of chronic inflammation (14) where local dysfunction acts to induce oncogenic mutations, genomic instability, and early tumor promotion and enhance angiogenesis (14). Tumors can directly disrupt and manipulate their local microenvironment by hijacking local cells and coercing them to provide varying services to help avoid the immune response or aid in other processes of tumor development by supplying cytokines, growth factors, and proteinases (15). Perhaps the best studied of these commandeered cells are the tumor-associated macrophage (TAM), the cancer-associated fibroblast (CAF), and the myeloid-derived suppressor cells (MDSCs).

The TAM is a pirated cell of the TME that provides local support to the tumor (16). In normal circumstances, macrophages act as a main defense against pathogens and a bridge between the innate and adaptive immune systems. As they become immersed in the evolving TME and are exposed to hypoxic conditions and tumor-derived factors, macrophages begin to change phenotype and functionality (17). This change may have serious implications for patient prognosis. A large array of clinical data indicate that the accumulation of TAMs within a tumor is associated with poorer outcomes (18). The production of TAMs was originally described as a phenotypic change of the M1 macrophage, which produces pro-inflammatory cytokines, to the M2 macrophage, known to produce anti-inflammatory and protumorigenic functions (19). This has been recently challenged in a mouse mammary cancer model that suggested TAMs are phenotypically and functionally different from M2 macrophages. The study further reports that TAMs more likely originate from CCR2⁺ inflammatory monocytes that depend on the notch signaling pathway for differentiation (20).

Cancer-associated fibroblasts are fibroblasts that have become activated by local factors within the TME (21). They have similar morphological properties to myofibroblasts (22) and assume the phenotype of a facilitator of tissue repair by generating and releasing growth factors and regulating inflammation and immunity (23). CAFs, unlike normal physiologically activated fibroblasts, are constitutively active and neither revert to a normal phenotype nor undergo apoptosis (24). They have been implicated in many aspects of tumor progression through varying mechanisms (25). For example, when CAFs isolated from human breast carcinoma were subcutaneously co-injected with a breast cancer cell line into an immunodeficient murine host, tumor proliferation was significantly increased in the presence of CAFs compared to normal mammary fibroblasts isolated from the same patient. Further experiments attributed this to the CAF’s ability to secrete stromal cell-derived factor-1 (SDF-1; also known as CXCL12) (26). It has also been shown that mice orthotopically co-xenografted with human pancreatic cancer cells and CAFs develop pancreatic tumors and metastases and that the CAF secretome stimulates the epithelial–mesenchymal transition (EMT) (27). CAFs also assist in metastasis by supplying transforming growth factor beta (TGF-β) to tumor cells, a multifunctional cytokine known to mediate the EMT (28), a process where cells lose epithelial markers and gain mesenchymal attributes allowing for a more mobile and migratory phenotype. TGF-β is also pro-angiogenic, and along with vascular endothelial growth factor (VEGF), platelet-derived growth factor (PDGF), fibroblast growth factor, and SDF-1 is released by CAFs to enhance endothelial cell proliferation and migration (25).

The MDSC is a more recently described cell type that becomes activated in pathological states and has potent immunosuppressive capacity (29). In some pathologies, MDSCs help protect the host organ from the harmful effects of excessive immune stimulation. For example, MDSCs were elevated in mice with antigen-induced autoimmune enterocolitis, and their presence lead to a reduction of disease symptoms (30). In cancer states, however, activation of MDSCs is an effective means of protection from immune-mediated killing (29) by inhibiting antigen-presenting dendritic cells, T and B cell proliferation, and natural killer cell cytotoxicity (31). Furthermore, MDSCs have also been implicated directly in tumor metastasis by helping to facilitate the EMT and to establish a distant premetastatic niche (PMN) (29).

The complex interactions displayed between cancer cells and the TME, as mentioned above, occur through a very complicated network of cellular communication. Many of these signaling pathways operate through direct cell-to-cell contact or using classical paracrine signaling loops of cytokines or growth factors with their receptors. However, more recently, EV shedding has emerged as another important mechanism of cellular cross-talk (7).

Extracellular vesicles are lipid bilayer-bound vehicles that are released from the cell membrane and carry nucleic acids (DNA, mRNA, and miRNA), proteins, and lipids to neighboring or distant cells (8). Although EVs were first described over 30 years ago as being released from reticulocytes (32), they have gained significant attention only recently as key factors in regulating both normal cell physiology and disease states. They now have been identified in nearly all eukaryotic cells (33) and prokaryotic cells (34) and have been isolated from most bodily fluids (8).

As shown in Figure 1, EVs are classified into two groups depending on their size, biogenesis, and method of release from the cell. Exosomes are 30–100 nm in diameter and are generated within large intracellular multivesicular bodies (35). They are released into the extracellular environment upon fusion with the plasma membrane.

Microvesicles (MVs) generally range from 100 to 1,000 nm and are formed when cell components travel to the plasma membrane to be released by membrane budding (36). Due to an incomplete understanding of exosome and MV biogenesis, and inconsistent methods of purification, the two terms are sometimes used interchangeably within the literature. The classical protocols for purification such as ultracentrifugation, density gradient centrifugation, and newer commercially available kits have been shown to co-isolate MVs and exosomes (37) as well as protein aggregates and other non-EV biomolecules that may interfere with EV specificity (38). Since current isolation methods are not yet standardized, it becomes difficult to assign specific functions to exosomes or MVs independently and why they are both included under the broad classification of EVs.

Extracellular vesicles exhibit wide-ranging roles in maintaining normal cellular and biological physiology. The EV lipid membrane protects its contents from enzyme degradation in bodily fluids making them ideal vessels to send material over a distance. They have been widely studied for their role in immune surveillance, blood coagulation, stem cell maintenance, tissue repair, and development (39). Since EVs play such a pivotal role in maintaining normal cell physiology, it is not surprising that their malfunctioning may lead to disease. EV involvement has been implicated in numerous pathological processes such as CCR5 receptor transport in HIV (40), beta amyloid transmission in Alzheimer’s disease (41), and the spread of prion disease to neighboring cells (42). Undoubtedly, however, the most heavily researched area of EV-associated pathology is their role in tumor development and chemotherapeutic resistance in cancer.

Extracellular vesicles represent a common method of communication within the local TME and distant sites. Although research pertaining to EVs derived from tumor cells monopolizes the literature, more evidence is emerging of the significant roles of stromal cell-derived EVs in cancer biology. EVs isolated from both cell types have been implicated in various steps of tumor growth and development including cell proliferation, angiogenesis, migration, invasion, metastasis, immunoediting, and drug resistance. For a comprehensive list of the molecules involved in tumorigenesis carried by EVs from both cancer and non-cancer cells, which are mentioned in this article, see Table 1.

The early detection of cancer is perhaps one of the greatest reasons for a favorable survival (108). Therefore, it is important to investigate novel tools for earlier cancer identification. Since the contents of EVs reflect the cell from which they originate, detection of specific EVs in bodily fluids may have diagnostic potential (109). EVs have been successfully isolated from blood, urine, ascites, cerebro-spinal fluid, amniotic fluid, semen, saliva, and bile (8), and many display a cancer-specific signature that allows for easy detection. For example, EVs isolated from the ascites of colorectal cancer patients contain enriched levels of caludin-3 protein compared to non-pathologic samples (110). Patients with esophageal SCC have serum-isolated EVs enriched with miRNA-21 (111). The specific EV contents also represent a snapshot of the tumor cell status at the time of release. Therefore, they may help in the development of personalized therapeutics or even monitor the tumor’s response to chemotherapy.

Ovarian cancer is known to present with vague and non-specific symptoms. This unfortunate circumstance accounts for the majority of women becoming diagnosed with advanced disease and is why the average 5-year survival is less than 50% (3). Therefore, screening techniques that identify specific cancer biomarkers may be a very useful tool for early detection and a subsequent drop in mortality. One group demonstrated that higher concentrations of an array of miRNAs including miR-21 and miR-141 within EVs were also present in ovarian tumor tissue. Furthermore, higher levels of these miRNAs were associated with more advanced disease (112). Since then, studies have identified other ovarian cancer biomarkers including epithelial cell adhesion molecule and CD24 (113). Efforts are now being made to find practical and cost-effective methods for detecting cancer-specific EVs in patient blood or other bodily fluids.

Scalpel-free biopsies, also known as liquid biopsies, are novel methods for cancer detection. These techniques detect free-floating cancer cells, circulating tumor DNA, or EVs within bodily fluids. There have been varying antibody-based methodologies tested for analyzing EV proteins from bodily fluids without an EV isolation step, including flow cytometry, EV protein microarray, diagnostic magnetic resonance, and nanoplasmonic sensing technology (114). Although still far from standard practice, some companies have already produced commercial kits for detecting lung and prostate cancer where specific EVs are isolated and analyzed using chromatin immunoprecipitation assays (115). In colorectal cancer, an assay called “ExoScreen” can detect cancer-specific circulating double-positive (CD146/CD9) EVs using photosensitive-beads from only 5 µl of patient serum (116).

Given the growing evidence of EV involvement in cancer pathogenesis, it seems intuitive to explore translational approaches that lead to their inhibition. Current studies are utilizing different techniques to inhibit vesicle formation, release, and cell uptake as well as blocking specific components of the EV. The drug amiloride has been shown in vivo to block secretion of tumor-derived EVs that contain membrane-associated heat shock protein 72 (HSP72) (117). HSP72 is constitutively expressed in many cancers and is associated with a poor prognosis (118). Furthermore, amiloride was shown to inhibit ceramide, an important mediator of EV biogenesis. Another drug, diannexin, inhibits phosphatidylserine, a regulator of cell adhesion, and EV endocytosis (119). Rab27 is a protein demonstrated to have a significant role in EV secretion (120). In highly metastatic mouse models of both melanoma and breast cancer, knocking down Rab27 led to a significant reduction of tumor EV production, primary tumor size, and metastasis (121, 122). However, since EVs are also essential participants in normal cell physiology, better techniques are required to distinguish and target pathological versus physiological EVs.

As mentioned previously, EVs have been used as cancer vaccines by carrying and providing tumor-specific antigens to immune cells, which prime the immune system and create a powerful immunological response against the tumor. Some of the first phase I clinical trials applying this methodology took place in melanoma and non-small-cell lung carcinoma where dendritic cell-derived EVs were loaded with MHC/tumor antigen and delivered back to patients (86, 87). These studies demonstrated that cancer vaccines are both feasible in creation and safe for administration. Later, EVs from the malignant ascites of colorectal cancer patients were isolated, mixed with specific cytokines, and administered back to the patient as a subcutaneous immunization. The investigators reported that combining ascites EVs with granulocyte-macrophage colony-stimulating factor induced specific antitumor cytotoxic T-lymphocyte activation (85). Furthermore, in a phase II clinical trial including patients with advanced small-cell-lung carcinoma, the administration of dendritic cell-derived EVs caused an increase in natural killer cell activity and longer progression-free survival for patients with low initial expression of natural cytotoxicity receptor NKp30 (88).

The realization that EVs are efficient vehicles for cell-to-cell communication has subsequently given rise to investigations of their use as a method for drug delivery. EVs display many potential advantages over current approaches. They are stable in serum, have specific cell-targeting capabilities, can overcome natural barriers such as the immune system or the blood–brain barrier, and can deliver molecules such as miRNAs or siRNAs that are readily degraded in the serum (38). Multiple methods have been reported to successfully load EVs with a desired drug. Hydrophobic drugs have been demonstrated to integrate with EVs successfully by simply mixing and allowing the drug to pass through the EV lipid bilayer membrane (123). The loading of hydrophilic drugs has proven to be more challenging, but still possible by methods including electroporation, sonication, saponin-mediated permeabilization, and freeze–thaw cycles (124). Perhaps the most challenging aspect of EV-mediated drug deliver is the efficient targeting of specific cell types. Some groups have used transfection-based approaches to encourage cells to express organ-specific ligands or receptors that are loaded into EVs, released from the cell, and then isolated and collected for successive drug loading (124–126). Other groups are experimenting with iron oxide nanoparticles in combination with a drug within EVs to target specific areas of the body by the application of a magnetic field gradient (127).

Currently, there is a rich source of data that creates a powerful case for the involvement of EVs in most, if not all, aspects of tumor development and progression. The intimate association of EVs with cancer has exciting implications for both cancer diagnostics and therapeutics. However, our general knowledge of EVs and EV-mediated processes remains in its infancy compared to other fields of cancer biology. Perhaps the most important hurdle that is restraining progress is the inability to consistently isolate EV subtypes, preventing investigators from comprehensively comparing and assigning to them specific functions. EV subtype uncertainty is demonstrated in much of the literature when investigators use the broad term “extracellular vesicle” as opposed to “microvesicle” or “exosome.” Isolation techniques cannot just focus on a single characteristic such as size or protein marker to discriminate among EV subtypes. Methods must use a combination of distinguishing features to better enrich the subpopulation, help define specific functions, and later to better categorize their role for therapeutics. Also, since EVs are key players in normal cell physiology, steps need to be taken to avoid interfering with normal processes. Despite current issues with EV isolation and subtyping, their role in cancer progression is dramatic and promises to become crucial in future advances in cancer diagnostics and therapeutics.

RS, GM, JH, and LL made substantial contributions to conception and design; RS performed literature search, analyzed data, and wrote the manuscript; GM, XZ, CS, JH, DM, and LL contributed to critically revise the manuscript; JH and LL supervised the study and final approval of the version to be submitted.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer, EP, and handling editor declared their shared affiliation, and the handling editor states that the process nevertheless met the standards of a fair and objective review.