Metastasis: an early event in cancer progression

Abstract

Purpose

Metastasis is the leading cause of death for a majority of cancer patients, and thus the need to understand the biology of metastasis becomes increasingly acute. When metastasis is initiated in tumor progression remains obscure. Better understanding of mechanisms regulating acquisition of metastatic ability in tumor cells will provide novel therapeutic targets and prevention of metastasis in clinics accompanied with the treatment of the primary tumor might be helpful in reducing metastasis-related mortality.

Methods

A literature search was performed in multiple electronic databases. Research papers from clinical reports to experimental studies on metastasis were analyzed.

Results

The article discusses tumor heterogeneity and genomic instability in the context of metastasis and tumor cell dissemination. And then we review biological mechanism of metastasis at an early stage in both intracellular (CSCs and CTCs) and extracellular (microenvironment) context. Finally, current development of anti-metastatic therapies is summarized.

Conclusions

Metastasis could be initiated at an early point of tumor progression. Therefore, early intervention on metastasis should be applied among cancer patients in clinical settings.

Introduction

It is well known that cancer without metastasis can often be treated successfully by surgery or local irradiation, and most cancer deaths, statistically as much as 90 %, are due to the development of metastases (Steeg 2006; Wang and Adjei 2015). Therefore, more should be learned of metastasis initiation on cellular and molecular levels, which would provide predictions about premium timings and targets of anti-metastatic therapies.

Metastasis is traditionally considered to be a relatively late process in tumor progression due to the prevailing clinical observation that metastasis is often diagnosed in primary tumors in later course of the diseases or with a diameter of more than 2 cm (Koscielny et al. 1984). However, removal of such primary tumors does not improve the chances of patient survival (Bedenne et al. 2007). Assessing the true patterns of metastasis only based on clinical observations such as tumor size becomes increasingly challenging because of the disrupted relationship between tumor size and metastasis spotted recently (Gonzalez-Angulo et al. 2009). Indeed, if metastatic regulation occurs late in progression as previously presumed, inhibitors of invasion should be able to prevent it. This, however, was not observed in several clinical trials testing inhibitors which mainly focus on later-stage primary tumors (Coussens et al. 2002). Most drug developments focus on short-term changes in primary tumor size in preclinical animal experiments, and clinical trials tend to enroll patients with advanced disease and established metastatic tumors (Bourzac 2014). Sadly, upon all the effort, mortality has improved only incrementally—a few months at best for most patients.

Emerging clinical evidence supports that some tumor cells may possess metastatic properties in the earlier stages of tumorigenesis (Husemann et al. 2008). Recent application of genomic approaches in determining molecular signatures also suggests that metastasis capacity may be acquired at the initial stages of tumor development (Bernards and Weinberg 2002). In this article, we first discuss clinically and biologically crucial aspects of tumor heterogeneity and genomic instability in the context of metastasis and tumor cell dissemination. We then review recent clinical and experimental evidences to discuss the relationship of cancer stem cells (CSCs), circulating tumor cells (CTCs), and tumor microenvironments with metastases and point to the fact that metastasis could be initiated at the same time or long before the first symptoms appear and the primary tumor is diagnosed. In both intracellular (CSCs and CTCs) and extracellular (microenvironment) context, we further provide our perspective on the possible biological mechanism of metastasis at an early stage. Finally, we review the current development of anti-metastatic therapies and urge that early intervention on metastasis should be applied among cancer patients in clinical settings.

Clinical evidence of early metastasis initiation

Human cancer progression is clinically determined by the tumor node metastasis (TNM) staging system (Harmer 1978; Tanase et al. 2015)—the extent of the disease at diagnosis is measured by tumor size (T stage), location of lymph nodes (N stage), and distant metastatic sites (M stage). When a total tumor mass reaches 1 kg, the organ usually fails, triggering several systemic processes and finally leading to death (Loberg et al. 2007). Therefore, tumor size has conventionally been thought not only as a fundamental criterion during diagnosis and treatment decisions but also as a critical determinant of clinical outcome. The therapy decisions are easily made due to the well-known notion that there is a good correlation between tumor size and grade: Larger tumors are generally of high grade. Also, biologically, tumor volume seems to be more relevant because it reflects the number of cancer cells; for every doubling of size the number of cancer cells increases eightfold (Klein and Hölzel 2006). The 15-year survival prognosis also suggest that for survival rate of patients with T1N0M0 breast cancer (tumor <2 cm, no lymph node and no overt distant metastasis) is 90 %, whereas for patients with T2N0M0 breast cancer (tumor diameter 2–5 cm, no lymph node and no overt distant metastasis), it is only 70 % (Koppelmans et al. 2014).

However, as we understand more and more about tumor biology and as the heterogeneity and complex ingredients of small-size tumors continue to be unveiled, the premise that positive correlation of tumor size and metastasis has come to the forefront. Metastasis was found at clinical presentation in early-stage cancer (T1M1 or T2M1) and developed from small tumors. Epidemiological analysis on various cancer patients further showed that metastasis might be initiated already 5–7 years before diagnosis of the primary tumor (Engel et al. 2003; Lim et al. 2012; Popiolek et al. 2013), suggesting that cancer cells capable to metastasize do not necessarily develop within large tumors. The clinical data of first to secondary metastasis also indicate that the relationship of tumor size and metastasis is not straightforward. Portal vein tumor thrombus (PVTT) is a clinical complication indicating invasive and early metastatic behavior in hepatocellular carcinoma (HCC). Previously, distinct PVTT (dPVTT), one type of PVTTs in patients with small HCC, was observed to be distant from liver parenchyma tumor nodules (Shuqun et al. 2007; Guo et al. 2010). Despite their small size, patients with dPVTT display high metastatic rate and poor survival. In another type of cancer, among T1 and T2 breast cancer patients, 5 % are present with metastasis at diagnosis (Zimmer and Steeg 2015). Furthermore, metastasis is even witnessed in the absence of detectable primary tumors (cancer of unknown primary), which ranks among the 10 most frequent cancer diagnoses (Pavlidis et al. 2015) and accounts for 5–10 % of diagnoses in Europe and the USA. The proportion of US patients in each category for the years 1988–2001 showed that of the four most prevalent cancers, about 10 % of breast and prostate cancer patients, 20 % of colorectal cancer patients. and 40 % of lung cancer patients had detectable distant metastases at diagnosis (Steeg 2006).

In summary, these findings hold important implications for understanding of metastasis initiation because they imply that development of metastasis could be simultaneous with that of primary tumors. These observations have also prompted a reconsideration of how, where, and when cancer cells acquire metastatic potential and have raised the possibility that cells with this potential may not be as rare in primary tumors as was originally believed. In light of these observations and the multilevel heterogeneity of systemic cancer in particular, focusing solely on observable criterion and corresponding traditional anti-metastatic therapies is unlikely to be the most promising approach.

Heterogeneity and genomic instability of metastases and disseminated tumor cells

Primary tumors are heterogeneous and contain numerous subpopulations of cells with genetic alterations. The intrinsic genomic instability of cancer cells increases the frequency of alterations necessary to acquire metastatic capacity. The genomic instability and heterogeneity of tumor cells are apparent in the chromosomal gains, losses, and rearrangements associated with cancer. Genomic stability could be interrupted by aberrant cell cycle progression, telomeric crisis (i.e., chromosomal instability), inactivation of DNA repair genes, and altered epigenetic control mechanisms (Schmidt and Efferth 2016). These oncogenic events creating genomic instability may also contribute to the evolution of tumors to the metastatic state. The prevailing clonal selection model of metastasis contends that the acquisition of metastatic ability is the final step in tumor development (Wu et al. 2012). Recent advances in the molecular profiling of cancer using genomic-level approaches challenges the model where acquisition of metastasis is considered to be the last steps of tumor progression. One study of mammary cancer showed that metastatic mammary epithelial cells were detected in the lungs of mice carrying the PyMT oncogene. These cells were observed early in breast tumorigenesis (or at a premalignant stage), suggesting the possession of disseminating and metastatic capacity even before full malignant transformation (Weng et al. 2012). Gene expression studies of van’t Veer et al. also revealed that large segments of the cancer cell population contained traits that predisposed these tumors to metastasis (Van’t Veer et al. 2002). Moreover, according to the widely held model of metastasis, rare subpopulations of cells within the primary tumor can acquire advantageous genetic alterations over time, implying that metastatic tumors should be genetically similar to the bulk of primary tumor cells. However, substantially different patterns of allelic losses, indicative of a high degree of genetic divergence, have been reported between primary tumors and metastases in prostate cancers (Cheng et al. 1999) and in breast cancers (Kuukasjarvi et al. 1997). The extent of genome diversity is also radical in many solid epithelial cancers including renal cell carcinomas (Bissig et al. 1999) and melanoma (Mitelman et al. 2007). These studies now lend their support to the notion that metastatic spread can be an early event during tumor progression; thus, primary and metastatic tumors might become quite genetically distinct as they evolve.

The metastatic ability of tumor cells allows them to surmount physical boundaries, disseminate, and colonize a distant organ. The stage at which individual cells leave the primary tumor is unclear. Revelation of cancer heterogeneity is not complete without analysis of disseminated tumor cells (DTCs), because they provide important clues regarding the relationship between primary tumors and metastatic sites. Analysis of single disseminated tumor cells in breast cancer demonstrated that disseminated tumor cells could be detected both in patients and in mouse models of cancer at very early, pre-invasive stages. An unexpectedly high genetic divergence has been witnessed in minimal residual cancer, particularly at the level of chromosomal imbalances (Husemann et al. 2008). Also in esophageal cancer, DTCs of lymph node and bone marrow metastases showed substantial divergence from primary tumors and among each other, suggesting parallel evolution induced by microenvironment and underlying the significance of divergent evolution between primary tumors and metastatic sites (Stoecklein et al. 2008). When more than one cell was analyzed, sibling cells from an individual patient rarely shared multiple chromosomal aberrations, but usually possessed distinctive genomes. CGH analysis of early disseminated tumor cells from patients without overt metastases showed that cells contained few chromosomal aberrations (Klein et al. 2002); further analysis unveiled the neoplastic origin of these cells and suggested early events in tumor progression (Schmidt-Kittler et al. 2003). DTCs from bone marrow and lymph nodes have also been shown to harbor characteristically different chromosomal aberrations (Stoecklein et al. 2008). All these data confirm that genome diversity is evident in early systemic cancer, which is inconsistent with metastatic spread being a late event. And genomic aspects of metastatic potential could be hardwired into cancer cells long before previously assumed.

Cancer stem cells: founder cells with metastasis capacity

CSCs have now been identified in a variety of solid tumors and hematological system cancers such as melanoma, breast cancer and acute myeloid leukemia (AML) (Lu et al. 2014; Emlet et al. 2014; Todaro et al. 2014; Lu et al. 2014; Al-Assar et al. 2014; Sarvi et al. 2014; Govaert et al. 2014; Nishimura 2014; Shlush et al. 2014) through xenotransplantation, and sphere formation assay. Many investigators have successfully established a correlation between the presence of CSCs in the primary tumor and increased metastasis incidence. In addition, a few studies have addressed the potential role of CSCs as cells with origin of distant metastases experimentally. Analogous to normal stem cells, CSCs also have several important “stem” properties (Dontu et al. 2003; Clevers 2015). These traits give CSCs the ability to renew themselves, to last a lifetime, to be resilient to electromagnetic and chemical exposure and to be able to slumber for prolonged periods of time and to colonize other parts of the body, which enables them to propagate long term or to sustain the disease. Among the most ominous properties of the subpopulation of these malignant cancer cells is their capacity to metastasize. Recent data in pancreatic cancer have demonstrated that a higher fraction of CXCR4⁺ cancer stem cells in primary pancreatic tumors displayed increased migratory activity in vitro and correlated with metastasis in follow-ups among these patients. In colon cancer, Pang et al. identified CD26 as a marker of migratory and distant metastasis-causing CSCs in colon cancer. Significantly, none of the patients with CD26⁻ cells in their primary tumors developed metastases (Pang et al. 2010). These results prove the existence of migratory CSCs, as originally proposed by Brabletz et al. (2005), and also provide direct evidence of the initiating role of CSCs in metastasis formation. Similar findings were reported in other cancer types such as breast cancer and HCC (Yang et al. 2008; Croker et al. 2009). Although all cancerous cells were able to establish themselves in certain organs, only a specific subpopulation of CSC-like cells were demonstrated to be capable of growing into large metastases elsewhere.

Although CSCs could contribute to the subsequent multiple steps of the invasion–metastasis cascade, the regulatory mechanisms of their metastatic ability remain relatively elusive. A set of molecules and pathways recently being identified to regulate both stem cell features such as adhesion, migration and protection, and cancer metastasis shed light on the possible mechanisms of how CSCs acquire the metastasis ability. Many of the factors known to govern normal stem cell features are also critical mediators of cancer metastasis. Examples include G-CSF and signaling through the laminin receptor found in the cancer of hematopoietic system (Papayannopoulou 2004), matrix metalloproteinase-9 (MMP-9) (a molecule regulates stem cell homing and migration) in prostate cancer (Chinni et al. 2006), aldehyde dehydrogenase (ALDH) (an enzyme involved in stem cell self-protection) in breast cancer (Croker et al. 2009; Charafe-Jauffret et al. 2010), and SDF-1–CXCR4 axis (master regulator of stem cell trafficking) in multiple cancer types (Kucia et al. 2005). Recent efforts also focus on the relationship between “stemness” markers including Oct4, Nanog, and Sox2 and metastatic capacity, and investigate dysregulation of similar self-renewal pathways such as Wnt, Hedgehog, and Notch (Takebe et al. 2011). Intercellular adhesion molecule 1 (ICAM-1), a molecule believed to be involved in stem cell adhesion, was also demonstrated to be a marker of HCC stem cells in humans and mice. Moreover, ICAM-1 expression is regulated by Nanog, a stemness factor. And inhibitors of ICAM-1 substantially slow tumor formation and metastasis in mice (Liu et al. 2013). These data point out that these molecules constitute a complex network of cellular interactions that facilitate the initiation of the premetastasis niche by the primary tumor and CSCs may be born with metastatic potential. Therefore, identification of novel markers and early targeting of the ‘beating heart’ of the tumor, CSCs, will be crucial to prevent metastasis.

Circulating tumor cells (CTCs): transferred cells with migrating capacity

CTCs are cell set loose by the primary tumor into the bloodstream and some of them will leave blood circulation and home into secondary organs. Obviously, they have accomplished most of the antecedent steps of metastasis. Therefore, CTCs that have been shed into the vasculatures and made their way to potential metastatic sites are of great interest in revealing mechanisms of the early occurrence of metastasis and could be key to cancer diagnosis or treatment (Hayes and Smerage 2009; Fehm et al. 2006). The concept of “CTC traffic” has been initially proposed whereby disseminated cells migrate between the primary tumor, bone marrow, and metastases site(s) in patients with advanced-stage cancer (Kim et al. 2009; Pantel and Brakenhoff 2004). Data on both cellular and molecular levels recently, however, point to a different scenario.

A variety of studies show that tumor cells are found to disseminate throughout the body in the early development of primary tumors before any sign of metastasis, or even before they become clinically detectable (Klein 2009). The disseminated tumor cells remained dormant for varying periods of time depending on the tissue, resulting in staggered metastatic outgrowth. With advances in technology, it has become evident that 20–40 % of patients with various epithelial tumors (e.g., carcinomas of the breast, prostate, colon, or lung) harbor occult metastatic cells in their bone marrow even in the absence of any lymph node metastases (stage N0) and clinical signs of overt distant metastases (stage M0) (Pantel et al. 2003). Particularly in breast cancer, the percentage of CTCs spotted in early-stage breast cancer patients ranging from 35.0 to 44.1 %, a relatively large proportion among the whole cohorts (Ignatiadis et al. 2007). Additionally, in a meta analysis of several studies on disseminated cells in breast cancer of more than 4700 patients, Braun and coworkers observed an increase in positive bone marrow samples from 22 % in stage T1 to 34 % in stage T3, whose gap was less significant than the authors hypothesized (2T1 < T3) (Braun et al. 2005). This study indicates that, unlike previously assumed, large tumors in relation to small tumors seed far more tumor cells to bone marrow. Thus, the poorer prognosis of patients with later-stage breast cancers might not be explained by increased disseminated cell numbers. In other types of cancer, similar concerns are addressed in both patients and transgenic mouse models. Recently, Rhim et al. (Rhim et al. 2012) developed a lineage-labeling system to detect and isolate cells of pancreatic epithelial origin during stochastic tumor progression (Rhim et al. 2012). During pancreatic tumor formation, tagged cells, mostly CTCs, were observed to invade and enter the bloodstream unexpectedly early, before overt malignancy could be detected by routine histological analysis. This is further backed by findings that patients with small (T1 stage) tumors harbor similar numbers of disseminated cancer cells compared to patients with late-stage tumors (T3–T4) present in the cohort of 607 patients. Also, in a melanoma mouse model—a model of metastasis, established by Eyles et al. (Eyles et al. 2010) tumor cell dissemination to distant organs occurs surprisingly early during the development of the primary tumor. In mammary tumor virus (MMTv)–Erbb2 and MMTv–PyMT (polyoma middle T antigen) transgenic mouse models with spontaneously arising mammary cancers, CTCs were also spotted in bone marrow and lung tissue before the primary tumor became morphologically invasive (Podsypanina et al. 2008; Huang et al. 2008). Furthermore, studies of breast and lung cancer even showed occult metastatic seeding may be mediated by CTCs that would not meet a standard definition of cancer (Husemann et al. 2008; Podsypanina et al. 2008).

Accumulating evidence on molecular level also suggest that dissemination of CTCs occurs in early stage of tumor progression. Genetic analysis in various cancer types revealed CTCs have significantly fewer genetic aberrations than primary tumor cells (Stoecklein et al. 2008; Schmidt-Kittler et al. 2003; Weckermann et al. 2009). In breast cancer, about 50 % of cytokeratin-positive CTCs had relatively normal karyograms, whereas all matched primary tumor karyograms were abnormal (Schardt et al. 2005). Although they seemed karyotypically normal, these CTCs were shown to contain small, subchromosomal deletions that are typical for breast cancer, proving their malignant origin and indicating that they disseminated before genome-wide instability was acquired. All these data outlined above indicate that dissemination of cells occurs early in the morphological development of cancer, at least in certain types of cancer and might not be a late event in cancer progression.

Microenvironment: indispensable factor for facilitating initial metastasis niche

Recent preclinical and clinical research has provided evidence that underlying genetic alterations of tumor cells and interactions within the tumor microenvironment are equally important to metastasis (Paget 1889). Moreover, the metastatic potential is not simply an inherent trait of cancer cells, but is substantially modified by the microenvironment, is clear (Alderton 2015). Studies of Bissell et al. (2005) showed that even fully malignant breast cancer cells could be reverted to a normal phenotype by exposing them to a non-permissive stroma; conversely, cells could acquire malignant potential at sites of wounds or irradiated tissue (Mueller and Fusenig 2004).

Several studies now also indicate that the “preparation of fertile soil” may begin earlier than previously thought. One of the studies showed that malignant transformation of mammary epithelial cells may occur in ectopic microenvironments such as the premetastatic lung in the absence of detectable metastatic progression (Podsypanina et al. 2008). The premetastatic environment could facilitate their lodging at secondary sites. With the help of some proteins and molecules, tumor cells with metastatic potential interact synergistically with their surrounding microenvironment, forming the metastatic niche in situ (Furger et al. 2001; Kang et al. 2003). Other components within tumor microenvironment such as soluble factors are also important players. Some signaling pathways such as stromal cell-derived factor-1 (SDF-1α) and vascular endothelial growth factor (VEGF) are important prerequisites for the initiation and progression of metastasis (Landskron et al. 2014). Such factors secreted from the primary tumor may prepare the metastatic niche and foster early cancerous colonies. In another phenomenon, termed angiogenic dormancy, there is a balance of proliferation and apoptosis of tumor cells with metastatic potential and neo-angiogenesis that results in micrometastases which do not progress further (Sosa et al. 2014; Shaked et al. 2014). However, when perturbations in the microenvironment occur, their re-activation will be able to form a clinically relevant metastasis (Kaplan et al. 2005). In fact, it might not be the elimination of disseminated cells that leads to a better outcome because tumor cells disseminated much earlier than previously thought. Thus, large primary cancers in patients may support metastasis not only by seeding more cancer cells, but also by providing unknown systemically acting factors that stimulate colonization of previously disseminated tumor cells at the ectopic site. All these results strongly point to the fact that although the genetic makeup of premetastatic cells is undoubtedly pivotal in determining metastasis, microenvironmental issues are also important in permitting malignant cells to realize their metastatic potential.

Tumor niches are usually made of non-malignant stromal cells, soluble factors, vascular networks, nutrients and metabolic components, and the structural extracellular matrix (ECM) (Wood et al. 2014). Interactions with and between cells in the premetastatic niche have generally been assumed to occur through cell–cell contact or the release of soluble tumor-derived factors. Recent advances have also demonstrated that extracellular vesicles (EVs) including microvesicles and exosomes play an important role in mediating premetastatic changes in putative sites of metastasis formation. EVs from tumor cells with high metastatic potential have been shown to carry significantly different cargoes than EVs from poorly metastatic cells, including some specific proteins and noncoding RNAs (ncRNAs) (Peinado et al. 2012; Hood et al. 2011; Rana et al. 2013), which are involved in metastatic potential and premetastatic niche formation. Dysregulation of miRNAs has been implicated in tumorigenesis and metastasis. Some studies in metastasis primarily focused on the impact that miRNAs have on the intrinsic properties of cancer cells. Forkhead box M1 (FOXM1), a proliferation-associated transcription factor, is shown to be also intimately involved in tumorigenesis of HCC under hypoxia environment or stimulated by certain inflammatory factors such as HBV. Meanwhile, FOXM1 is found to positively regulates the expression of miR-135a and that the expression of metastasis suppressor 1 (MTSS1) is down-regulated by miR-135a, which can promote PVTT initiation and progression. These results indicate that change of environmental factors could initiate metastasis simultaneously with tumorigenesis (Liu et al. 2012). Recent reports reveal that miRNAs are also actively involved in the formation and function of different microenvironments during tumor dissemination. The miR-200 family was reported to regulate epithelial–mesenchymal transition (EMT), one of the migratory programs and a crucial process for the establishment and maintenance of cancer cell stemness in microenvironment (Gregory et al. 2008; Korpal et al. 2008; Humphries and Yang 2015). Moreover, accumulating evidence has pointed out that miRNAs can regulate metastasis environment in multiple ways (Zhang et al. 2013; Gao et al. 2013). Through modulation of the tumor microenvironment, those miRNAs can regulate cancer cell interactions and their metastatic potential. For example, HBV could result in genesis of HCC and up-regulation of miR-143 expression. And the latter promotes cancer cell invasion/migration and metastasis by repressing expression of a fibronectin protein, FNDC3B. One of the most interesting parts of the study is that HBV triggers metastasis initiation at the same time with HCC genesis via miR-143 (Zhang et al. 2009). Apart from miRNAs, another ncRNA, lncRNAs which lack protein-coding capacity have also been considered to be one of the critical factors in the metastatic cascade recently (Gutschner and Diederichs 2012). HOX transcript anti-sense RNA (HOTAIR) is the first example of lncRNA that was found regulating distant genes. Gupta et al. recently revealed an important role for HOTAIR in breast cancer metastasis. HOTAIR is highly induced in metastatic breast cancer, and a high level of HOTAIR expression in primary breast tumors is a powerful predictor of eventual metastasis and mortality (Gutschner and Diederichs 2012). Other lncRNAs such as metastasis-associated lung adenocarcinoma transcript 1(MALAT1) were also reported to be a prognostic marker for metastasis and patient survival in colorectal cancer and HCC (Ji et al. 2003; Kogo et al. 2011). Moreover, lncRNAs were also found to promote the formation of an early metastasis niche. Hypoxic microenvironments have a critical role in metastasis, especially in the early formation of the premetastatic environment of various cancers (Ling et al. 2015). It has been demonstrated that down-regulation of intronic transcript 1 (NPTN) help create a hypoxic microenvironment through histone deacetylase 3 (HDAC3) activation and NF90 protein ubiquitination and subsequent degradation, resulting in hypoxia-induced tumor cell invasion and metastasis (Ling et al. 2015). All the above data suggest that components within the tumor microenvironments are all essential to metastasis and that these molecules initiate preparation and regulation of “metastatic soil” simultaneously during tumorigenesis. Timely interrupting reciprocal interactions between tumor cells and the primary or foreign microenvironment could be promising therapeutic efforts in blocking metastatic colonization.

Development of anti-metastatic strategies

As most metastatic cancers are inoperable, treatments using chemotherapy or/and radiotherapy are currently the common option, mainly eliminating rapidly proliferating cells via inhibition of DNA replication, DNA repair, or the cell cycle, to slow tumor growth and relieve metastasis-associated morbidity. Recently, targeted treatments aiming at interrupting oncogenic and tumor-specific pathways, eradicating DTCs (or keeping them dormant), and depriving microenvironmental facilitators have become major components of anti-metastasis therapy.

Although microarray data showed that not much is shared between primary and metastatic sites, there are still several pathways and molecules functioning in both tumorigenesis and metastasis. For example, rigorous research has shown that oncogenic BRAF signaling, previously thought to have an important role in the pathogenesis of malignant melanoma (Shen et al. 2016; Flaherty et al. 2010), also show dramatic efficacy in the treatment of metastatic malignant melanoma in initial clinical trials (Bollag et al. 2010). Drugs based on oncogenic signaling pathways such as imatinib, trastuzumaba, and erlotinib have been successfully developed and put in trials in treating metastatic diseases (Joensuu et al. 2012; Qin et al. 2013; Rosell et al. 2012). There are also emerging strategies of using antagonists to elevate expressions of metastasis suppressor genes (Resovi et al. 2014; Gao et al. 2012). One of them is Atrasentan, an ET-1 receptor antagonist in bladder cancer. An adjuvant clinical trial of Atrasentan is currently being planned for patients with locally advanced bladder cancer. A preclinical study on bladder carcinoma cell lines identified ET-1 as a correlate of low RhoGDI2 expression, and Atrasentan inhibited the pulmonary metastasis of a low RhoGDI2 expressing bladder cancer cell line by 80 % (Piergentili et al. 2014).

Under the present diagnostic system, patients with cancer often already have a substantial number of DTCs or even undetected micrometastasis upon initial diagnosis (Pantel et al. 2008). Therefore, anti-metastasis therapies should be capable of targeting those cells that have escaped from the primary tumors. There are currently two approaches: (1). eradicating DTCs and (2). keeping DTCs dormant. In the first approach, straight strategies include targeting DTC-specific survival pathways such as the c-Src, VCAM1, and unfolded protein response (UPR) pathways (Zhang et al. 2009; Lu et al. 2011; Sosa et al. 2011). Indirect strategies usually applying niche-targeted approach use antagonists as receptors for the niche constituents to mobilize DTCs out of a dormant niche so that they might be sensitive to conventional chemotherapy (Becker 2012). The second approach is most alluring, because it could obviate the need for chemotherapy and the associated side effects. One of the ways is to induce expression of dormancy-initiating factors in DTCs. For instance, the stress-activated protein kinase p38 and homeobox transcription factor D10 (HOXD10) are known drivers of dormancy in different cancers (Aguirre-Ghiso et al. 2001; Chen et al. 2009). Therefore, strategies aiming at sustaining p38 or HOXD10 in dormant DTCs could theoretically result in maintenance of DTCs dormancy. Other ways includes chronic suppression of activating pathways of DTCs (Beliveau et al. 2010) and providing constituents of the dormant niche to permanently rein DTC growth (Catena et al. 2013).

Cellular microenvironment of the secondary site has a crucial role in facilitating metastasis. Often, facilitators in the microenvironment help DTCs successfully colonize a secondary organ and alter metastatic cells in ways that render them resistant to cell-autonomous therapies. Anti-angiogenic therapy is a prime example of targeting microenvironmental facilitators. For example, in colorectal, ovarian, and non-small cell lung cancers, treatment outcome of bevacizumab, an antibody targeting vascular endothelial growth factor, in combination with chemotherapy, yielded progressive results (Jackisch et al. 2015). Other examples include combretastatin A4 (a vascular disrupting agent) (Jung et al. 2016) and other small molecules and antibodies that target angiogenic growth factor receptors such as PDGF-R, VEGFR-3, and HIF (Moreira et al. 2007).

Conclusion and perspective

The data we summarized above indicate that regulation and spread of metastasis occurs early, which reinforces the need to dissect metastasis process in further detail in order to help prevent metastasis (Fig. 1a). Currently, treatment for metastasis is always delayed and not well appreciated. Therefore, prevention and treatment of metastasis should be accompanied with the treatment of primary tumor, right after diagnosis being confirmed (Fig. 1b). Discovery of early indicators of metastasis is also urgently required. Although all these studies provide thrilling insights into the molecular mechanisms of metastasis in recent years, too many questions still remain to be answered in basic research. First, we need to identify the true cell contributor(s) of metastasis. CSCs are considered as a small subpopulation of tumor cells with “metastasis-initiating potential,” while CTCs are considered as the tumor cells with a “migrating ability” (Oskarsson et al. 2014). Whether either group or subpopulation of either group or cells with features of both groups contributes directly to metastasis remains unclear. In colon cancer, Dieter et al. (2011) demonstrate that extensive self-renewing long-term tumor-initiating cells (LT-TICs) are the only subpopulation of CSCs able to contribute to metastasis formation. Also, Balic et al. (2006) find that most early disseminated cancer cells detected in the bone marrow of breast cancer patients have a putative CSC phenotype. Therefore, more efforts should be made to identify the real subpopulation(s) of tumor cells that contribute to the metastatic-initiating process. Second, integrated strategies of targeting metastatic cells based on their molecular and cellular nature are needed. Technologies to identify and isolate CSCs and CTCs are staggered by various issues. Variable features of CSCs make it difficult for scientists to accurately target them (Hu et al. 2013). As for CTCs, EMT may play a key role in the ultimate generation of metastatic foci (Lamouille et al. 2014). Thus, epithelial-specific cell surface markers are in widespread use to detect CTCs in multiple cancers (Krebs et al. 2011; Cohen et al. 2006; Cristofanilli et al. 2004; Li et al. 2014; Nagrath et al. 2007). However, this represents two potential problems: One is the application of them are limited in epithelial tumors, and the other is because it is likely that carcinoma cells passing through a partial or complete EMT are no longer detectable by epithelial-specific antigens. Novel markers for detecting non-epithelial CTCs will help tackle the former problem, and enrichment and detection of CTCs (the so-called negative selection via depletion of hematopoietic cells using antibodies specific for CD45) may represent one way of circumventing the latter issue (Alix-Panabières and Pantel 2014). Third, we also need to learn more about how metastatic cells acquire metastatic potential and work to their metastatic capacity by analyzing phenotypic properties, hosting niches and signaling pathways of them and the related microenvironmental influences. By interrupting the premalignant niche and reconstructing normal microenvironment, we may be able to disrupt the acquisition of competence by these cells to colonize distant organs. Building on increasing evidence emerging from recent studies (Liu et al. 2011, 2013; Seton-Rogers 2015), it is likely that the link between ncRNAs and metastasis will find further experimental and clinical relevance. All these data might offer novel predictors and therapeutic targets for anti-metastasis treatment in the near future.

Fig. 1.

a Metastasis process and possible therapeutic targets. Metastatic tumor cells (light blue and purple) invade locally through surrounding extracellular matrix (mustard) and stoma cell layers (pink) (1), intravasate into blood vessels, survive the transport through the vasculature (2), extravasate into distant tissues and eventually generate clinically detectable metastasis (3). Potential targets for preventative metastasis therapy include CSCs (purple), CTCs (light blue) and microenvironment factors such as soluble factors (blue dot), neoangiogenesis (red) and ncRNAs (green). b Time point of metastasis therapy. Current metastasis therapy in clinics usually awaits until stage 3 and treatment outcomes are mostly gloomy. Initial treatments for cancer usually do not include selected therapies against metastasis. In fact, CSCs, and CTCs acquire or realize metastasis potential predominantly during stage 1 and stage 2. Formation of initial metastasis niche facilitated by certain environmental factors also mostly start during stage 1. Therefore, stage 1 and stage 2, especially stage 1, are better time points for effective metastasis treatment. Early prevention of metastasis might greatly improve prognosis among cancer patients

On the other hand, the development of effective anti-metastatic strategies has three potential barriers: (1) Since tumors are heterogeneous, it could be that the metastasis signature is represented by the bulk population, but not by every cell within that population. Some subpopulations within the tumor could express one of the metastasis signature genes, whereas others could express larger proportions of the signature genes. One of the solutions to tackle the problem is the use of a cocktail of multiple anti-metastatic agents targeting metastatic cancer cells and their related microenvironments with cytotoxic chemotherapy. The conduct of trials using drug combinations produced by multiple companies might also be key to the success of this effort. Hopefully, the combination of these anti-metastatic therapies could be a promising way for improving long-term survival among cancer patients. (2) Another potential factor is the reliance on primary tumor biology for metastasis drug development. Studies such as Bcl-xL and the insulin receptor substrate 2 indicate that metastasis can be modified in the absence of a change in the primary tumor (Martin et al. 2004; Nagle et al. 2004). Further preclinical studies report differential effects of drugs on primary and metastatic disease (Nasulewicz et al. 2004; Shannon et al. 2004). Therefore, more metastasis-specific signatures need to be identified. As for the making of anti-metastasis tactics, metastatic cancer cells and its surrounding microenvironment render metastasis sites stronger resistance to current treatment as compared to primary sites. We need to develop more specific and tailored therapeutic targets to prevent and treat metastasis. (3) Current therapeutic strategies have limited success in preventing and treating disease in patients with metastasis, because the reduction of tumor burden in patients with late-stage cancer is mostly palliative and rarely lead to complete and sustained remission. Even efforts are made during angiogenesis and colonization to form a detectable metastasis; they are still last steps in the development of metastasis. Treatment outcomes are looming. Thus, more clinical trials ought to be designed to evaluate anti-metastasis therapy in the setting of early-stage cancer, during which intravasation and extravasation of the circulatory system and colonization as an occult micrometastasis has occurred. This may represent a point in disease progression when metastasis could be preventable.

Cancer biologists and oncologists have long anticipated the opportunity to discuss cancer treatment strategies with patients in the context of a “chronic illness.” With careful surveillance and the timely intervention of metastasis, the leading cause of cancer-associated death, cancer patients could live peacefully with cancer and their disease may be managed as a chronic one. This requires a deeper understanding of metastasis and the design of more effective cancer therapies. In fact, metastasis, as an early event argues for a shift of future bench research to identify the cells with real metastatic potential and understand how tumor dormancy and metastatic outgrowth are regulated. The importance of early intervention on metastasis needs to be urgently recognized. And preventative treatment for metastasis should be applied significantly earlier than current practices.

Compliance with ethical standards

Conflict of interest

The authors have declared no conflicts of interest.

Ethical approval

This article does not contain any studies with human participants or animals performed by any of the authors.