Tumour–adipose tissue crosstalk: fuelling tumour metastasis by extracellular vesicles

Abstract

During metastasis, tumour cells must communicate with their microenvironment by secreted soluble factors and extracellular vesicles. Different stromal cell types (e.g. bone marrow–derived cells, endothelial cells and fibroblasts) influence the growth and progression of tumours. In recent years, interest has extended to other cell types in the tumour microenvironment such as adipocytes and adipose tissue–derived mesenchymal stem cells. Indeed, obesity is becoming pandemic in some developing countries and it is now considered to be a risk factor for cancer progression. However, the true impact of obesity on the metastatic behaviour of tumours is still not yet fully understood. In this ‘Perspective’ article, we will discuss the potential influence of obesity on tumour metastasis, mainly in melanoma, breast and ovarian cancer. We summarize the main mechanisms involved with special attention to the role of extracellular vesicles in this process. We envisage that besides having a direct impact on tumour cells, obesity systemically preconditions the tumour microenvironment for future metastasis by favouring the formation of pro-inflammatory niches.

This article is part of the discussion meeting issue ‘Extracellular vesicles and the tumour microenvironment’.

1. Introduction

Metastasis is still the major factor related to morbidity and mortality among cancer patients. For over a century, cancer biologists have intensely explored the mechanisms underlying the emergence and spread of tumour cells, yet how cancer cells acquire the competence to colonize distant organs remains a central question in cancer biology [1,2]. Although tumour cells are the driving force of metastasis, new findings suggest that host cells within the tumour microenvironment play a key role in guiding metastatic behaviour [3]. Thus, bidirectional communication between tumour cells and their microenvironment is crucial for tumour initiation and progression.

Within primary tumours, multiple stromal cell types are involved in the formation of a pro-tumourigenic microenvironment [4]. Tumour cells release several soluble factors (e.g. vascular endothelial growth factor A (VEGF-A), tumour necrosis factor alpha (TNFα) and transforming growth factor beta (TGFβ)) and pro-inflammatory chemoattractant proteins (e.g. S100A8 and S100A9) that mobilize populations of myeloid/haematopoietic cells to settle in distant organs and tissues [4]. These cells establish ‘pre-metastatic niches’ with a microenvironment that supports the incoming circulating cancer cells [5–7]. Secreted proteins, such as members of the Lysyl oxidase (LOX) family, participate in the formation of pre-metastatic niches (including LOX, LOX-like 2 (LOXL2) and LOX-like 4 (LOXL4)), remodelling the extracellular matrix (ECM) and favouring the adhesion of bone marrow–derived cells (BMDCs) [4–7]. Thus, the deposition of ECM proteins by activated stromal fibroblasts (e.g. fibronectin) also plays a crucial role in the attraction and homing of BMDCs to pre-metastatic niches [8].

In addition to soluble factors and secreted proteins, there has been increasing interest investigating the role of extracellular vesicles (EVs) in the formation of the pre-metastatic niche and in metastatic colonization [9–12]. EVs are classified into two main subtypes according to their size and origin: microvesicles and exosomes. Microvesicles (100–1000 nm) are a heterogeneous population of vesicles that bud directly from the cell membrane, whereas exosomes (30–100 nm) are derived from the luminal membranes of multivesicular bodies and they are secreted by cell membrane fusion events controlled mainly by Rab family proteins, specifically Rab27a and Rab27b [13]. EVs are involved in the horizontal transference of material between cells, modulating the phenotype and functional characteristics of the recipient cells [14,15]. Different routes of exosome uptake have been described, including plasma membrane fusion, endocytosis or phagocytosis, and even via receptor binding that may also activate specific signalling pathways in the recipient cell [16]. Owing to their long-term stability in the bloodstream and other body fluids, circulating EVs have emerged as potential biomarkers in a wide variety of diseases, including cancer [17]. However, isolation techniques and technical standardization remain a challenge when considering the use of these vesicles, and protein and lipid aggregates may also be isolated as contaminants in EV preparations. High concentrations of lipids and lipoproteins may particularly hinder the isolation of non-contaminated EV preparations in a context of obesity. Furthermore, identifying specific markers suitable to analyse different EV populations is still limited by our understanding of the physiological and pathological roles of EVs [18].

Tumour-secreted exosomes play a crucial role in pre-metastatic niche formation, enabling tumour cells to communicate with neighbouring and distant cells. Indeed, they are responsible for the transfer of proteins and genetic material from primary tumour cells that drive the recipients to acquire a pro-tumourigenic phenotype [9,15]. Thus, the uptake of tumour-secreted exosomes by stromal cells induces a reprogramming that contributes to the formation of the pre-metastatic niche. For example, the c-MET oncoprotein found in melanoma-derived exosomes reinforces the formation of the lung pre-metastatic niche and c-MET expression in bone marrow progenitor cells favours metastatic progression [11]. Similarly, CD105-positive microvesicles have been implicated in the expression of pro-angiogenic mRNAs and microRNAs in pre-metastatic lung niches in models of renal cancer [19]. It was also reported that the CD44 variant isoform CD44v6 is required for the assembly of a soluble matrix involved in pre-metastatic niche formation in lungs and lymph nodes in cooperation with tumour-secreted exosomes in pancreatic cancer models, reinforcing leucocyte, stromal and endothelial cell activation [20]. Furthermore, tumour-secreted exosomes influence metastatic organotropism, preparing pre-metastatic niches in the liver and lungs [10,12]. In particular, exosomal integrins, such as α6β4/α6β1 or αvβ5, are involved in the organotropic behaviour of several tumour types, inducing lung and liver metastasis, respectively [10]. Therefore, tumour-secreted exosomes help establish tumour-supportive microenvironments in distant organs prior to the arrival of tumour cells, favouring the survival and outgrowth of the latter. As such, the metastatic behaviour of tumour cells clearly depends on intrinsic and extrinsic factors, and the active crosstalk between tumour cells and their microenvironment is crucial to regulate metastatic cell behaviour. Accordingly, it has become clear that other environmental factors, such as obesity, are likely to influence cancer development and progression [21,22].

Obesity is considered a risk factor for the development of different types of cancer and this co-morbidity worsens the prognosis of cancer patients [21,23,24]. Epidemiological evidence links obesity to the risk of developing pancreatic, prostate, breast, colon, ovarian, endometrial, liver, kidney, oesophagus, gastric and gallbladder cancers, to name a few [21,22,25]. Although white adipose tissue has traditionally been considered as an energy store, nowadays it is considered to have a central role in both endocrine and metabolic processes [21]. Hence, it is necessary to understand how obesity and adipose tissue–related factors affect tumour development and progression. Little is known about the implication of obesity on metastasis and it is unclear whether, in addition to its impact on primary tumour development, it influences the metastatic behaviour of certain tumours. Do specific factors related to obesity (e.g. lipids, soluble factors, EVs) promote metastatic behaviour? Could obesity influence systemically organs for future metastasis by promoting the formation of pro-inflammatory niches? These and many other questions remain to be answered, and, thus, in this ‘perspective’ article we will discuss how obesity could contribute to the metastatic process.

2. The role of soluble factors in obesity-associated inflammation during tumour progression and metastasis

Adipose tissue is composed of adipocytes and non-adipocyte cells, including mesenchymal stem cells and macrophages. These cell types release a wide variety of molecules that enable them to communicate with other cells, and to play an active role in pathological processes such as breast and ovarian cancer [23,26–28]. Obesity induces chronic inflammation in adipose tissue, which is characterized by the secretion of pro-inflammatory cytokines and by the infiltration of macrophages, cells that facilitate tumour progression and metastasis [29]. To date, there are at least 4 factors thought to associate obesity with tumour progression: (i) enhanced secretion of pro-inflammatory molecules [30]; (ii) altered energy metabolism [31]; (iii) secretion of pro-angiogenic factors [32] and (iv) immune cell recruitment and macrophage polarization [33]. In this scenario, secreted EVs have largely been ignored and they have only begun to be studied in recent years. Thus, in this section we will focus our attention on soluble factors, later addressing the potential role of EVs in tumour metastasis.

In an obese individual, adipose tissue becomes dysfunctional and the profile of adipokines that are secreted changes, elevating the pro-inflammatory factors within the local microenvironment [29]. In a tumour setting, the release of these factors alters the tumour environment, promoting its growth and progression [30,34,35]. Therefore, it is plausible that a pro-inflammatory microenvironment, such as that associated with dysfunctional adipose tissue, would favour both primary tumour progression and metastatic behaviour (figure 1).

Figure 1.

Obesity-induced inflammation contributes to melanoma progression. (a) Schematic of how obesity induces chronic inflammation in adipose tissue during melanoma progression through the secretion of specific adipokines and pro-inflammatory cytokines. Obesity soluble factors act as local elements that mediate primary tumour growth and tumour infiltration by macrophages, which promotes metastasis through the secretion of specific factors (see text for more details). (b) Schematic representation of the main pro-inflammatory cytokines and angiogenic factors secreted by adipose tissue that influence primary breast cancer growth and metastasis in the lungs and lymph nodes (see text for more details). (c) In ovarian cancer, obesity influences metastasis to the omentum due to the increase of specific adipokines. Direct exposure of ovarian cancer cells to leptin increases tumour migration and invasion, concomitant with EMT (see text for more details). Cancer cell–related changes are shown in boxes, and the factors modulated are shown in yellow (adipocytes), brown (melanoma), grey (macrophages), green (lymph nodes), purple (breast cancer) and blue (ovarian cancer).

Recently, an association between the ingestion of a hypercaloric diet and the risk of developing lymph node metastasis and lymphangiogenesis in melanoma was brought to light [36]. In this model, mature adipocytes from epididymal adipose tissue promote the secretion of soluble pro-inflammatory factors by tumour cells co-cultured with macrophages, such as CCL2 and M-CSF, enhancing M2 macrophage polarization and cytokine secretion, as well as angiogenic and lymphangiogenic factors (figure 1a) [36]. Furthermore, tumour-associated adipocytes enhanced CCL19, CCL21 expression in lymph nodes and CCR7 expression in tumour cells, favouring lymph node metastasis (figure 1a) [36]. Additional evidence of crosstalk between melanoma cells and mature adipocytes was obtained by studying intra-tibial injection of melanoma cells [37]. Indeed, adipocytes co-cultured with melanoma cells induce the secretion of pro-inflammatory cytokines (interleukin [IL]-1β, IL-6) and chemoattractants (CXCL1, CXCL2, CXCL5) by melanoma cells (figure 1a). Moreover, mice fed with a high-fat diet (HFD) upregulate the secretion of osteopontin by macrophages and promote osteoclastogenesis, which in conjunction with IL-6, induces melanoma cell proliferation in the bone marrow (figure 1a) [37]. Adipocyte-secreted factors can also directly influence metastatic behaviour in tumour cells as adipocyte-conditioned medium induces the epithelial to mesenchymal transition (EMT), invasion and the migration of melanoma cells in vitro, in part, due to IL-6 secretion [38]. In this model, adipocyte-conditioned medium induces Snai1 upregulation and the loss of E-cadherin protein, and the metastatic suppressor kiss1 [38].

HFD-induced obesity favours the growth of primary melanomas in C57BL/6 J mice, increasing leptin and reducing adiponectin levels [39]. Besides these changes in typical obesity-related molecules, tumours from HFD mice express caveolin 1 (Cav-1) and fatty acid synthase (FASN) more intensely, and their increased phospho Akt (pAkt) levels are associated with rapid melanoma tumour growth (figure 1a) [39]. Treatment of these animals with orlistat (a FASN inhibitor), or by restricting caloric intake, slows down melanoma tumour growth, decreases fat mass and reduces FASN, Cav-1 and pAkt levels [40].

Like melanoma, a HFD induces the secretion of pro-inflammatory cytokines (monocyte chemotactic protein 1 (MCP1/CCL2)), plasminogen activator inhibitor-1 (PAI1), TNF-α) and angiogenic factors (hepatocyte growth factor (HGF), TIMP metallopeptidase inhibitor 1 (TIMP1), VEGF) in MMTV-PyMT transgenic mouse model of breast cancer, both in the tumour and in the plasma. These changes are correlated with enhanced tumour growth and lung metastasis (figure 1b) [41]. Furthermore, HFD-altered energy metabolism by upregulating leptin and decreasing adiponectin levels could promote tumour metastasis in conjunction with cytokines and angiogenic factors [41]. Co-culture of murine and human breast cancer tumour cells with commercially available mouse adipocytes and mature human adipocytes isolated from mammary adipose tissue also enhances their invasiveness in vitro and in vivo [42]. Mechanistically, co-culture of tumour cells with adipocytes induces protease overexpression, including matrix metalloproteinase-11 (MMP-11), and pro-inflammatory cytokines (e.g. IL-6, IL-1β), favouring the invasion of breast cancer tumour cells (figure 1b) [42]. In addition, the CC-chemokine ligand 5 (CCL5) secreted by human mammary adipocytes when co-cultured with the MDA-MB-231 cell line also enhances invasiveness. Importantly, in triple negative breast cancer patients, immunodetection of CCL5 in adipose tissue surrounding the tumour is correlated with lymph node and distant metastases, and decreased overall survival (OS) of patients, highlighting the impact of these cytokines on breast cancer progression [43].

Ovarian cancer cells spread quickly within the abdominal cavity to the omentum, one of the most common sites of ovarian cancer metastasis [27]. Initial metastatic seeding in the omentum was attributed to the secretion of IL-6, IL-8, TIMP1 and MCP1 by omental adipocytes, highlighting the importance of a tumour-supportive microenvironment in this model (figure 1c) [27]. Interestingly, in ex vivo adhesion experiments using peritoneal tissue, the adhesion of ovarian cancer cells is enhanced in the tissue derived from mice with diet-induced obesity. Intraperitoneal injection of ovarian cancer cells confirmed these results, with an increase in tumour burden in obese mice in vivo [44]. This process relies on enhanced vascularity, a diminished M1/M2 macrophage ratio and altered lipid regulatory factors (fatty acid binding protein 4 (FABP4), sterol regulatory element-binding proteins SREBPs), although the specific molecular mechanisms involved were not defined [44]. The exposure of ovarian cancer cells to leptin increases tumour cell migration and invasion due to the activation of JAK/STAT3, PI3/AKT and RhoA/ROCK signalling downstream of the leptin receptor (Ob-Rb). Such enhanced signalling upregulates markers of stemness and EMT [45]. Furthermore, Ob-Rb is strongly expressed in metastatic lesions than in primary tumours (figure 1c), associated with worse survival in overweight patients. Hence, leptin has an important role in promoting ovarian cancer metastasis in association with an obese phenotype.

Collectively, these data suggest that dysfunctional adipose tissue associated with obesity enhances metastatic behaviour in preclinical models of breast cancer, melanoma and ovarian cancer through common mechanisms (figure 1): (i) influencing the secretion of pro-inflammatory factors from tumour cells; (ii) promoting M2 macrophage polarization, which in turn affects the secretion of pro-angiogenic and lymphangiogenic factors; (iii) reinforcing tumour cell invasion, migration and EMT. However, the relevance of EVs in this setting is unclear and thus, it would be interesting to determine whether EVs and soluble factors act in a coordinated manner in the context of obesity-induced inflammation.

3. The role of extracellular vesicles in tumour–adipose tissue communication

EVs are crucial in the communication between tumours and their microenvironment [9]. They have a lipid bilayer that contains a molecular cargo defined by the cell of origin, comprising proteins, nucleic acids and other molecules [15]. EVs can modulate the physiology of the recipient cells in various ways, principally by inducing intracellular signalling in recipient cells and by conferring novel properties associated with the acquisition of new receptors, enzymes or even genetic material from the cell of origin [46]. Thus, EVs serve as a vehicle for the horizontal transfer of mRNAs, small RNAs, micro RNAs and proteins between different cell types within the tumour microenvironment [9,15,46]. Exosomal communication has emerged as a new and widespread mechanism that must be considered when studying tumour progression and metastatic dissemination [9]. This mechanism acquires even more importance when associated with metabolic alterations that generate a microenvironment supportive of tumour growth [31]. Indeed, EV-secreted miR-122 reprogrammes glucose metabolism by increasing nutrient availability in the pre-metastatic niche and promoting metastasis in breast cancer models [47]. Hence, metabolic changes could be horizontally transferred between cell types via secreted EVs.

Bidirectional communication of melanoma cells with the adipose tissue has been recently reported to reinforce the aggressive phenotype in melanoma [48]. Thus, melanoma cells take up exosomes secreted by surrounding fat cells that favour their migration and invasion. Adipocyte-derived exosomes drive cancer cells towards a more aggressive phenotype due to the transfer of enzymes implicated in fatty acid oxidation (FAO: figure 2a). Furthermore, this effect was reinforced in association with obesity due to the larger number of exosomes secreted and the stronger effect of each individual exosome [48]. As previously mentioned, Cav-1 is an important molecule in the control of melanoma cell proliferation in mouse models of obesity [39]. Interestingly, Cav-1 is expressed in exosomes secreted by human melanoma cells (in vitro) and in the plasma of mice engrafted with melanoma tumours, whereas it was not evident in cell extracts and exosomes from normal human cells [49]. A significant increase in circulating exosomes expressing Cav-1 can be observed in the plasma of melanoma patients, suggesting that Cav-1 may be an interesting biomarker in melanoma patients [49]. Nevertheless, as body mass index was not analysed in this study, further studies will be necessary to establish any link with obesity.

Figure 2.

Communication between adipose tissue and cancer cells through secreted vesicles. (a) In melanoma, adipocyte-derived exosomes enhance migration and invasion in response to the horizontal-transfer of functional enzymes implicated in FAO. (b) Exosomes from AD-MSCs enhance primary tumour growth and the migration of breast cancer cells. (c) Ovarian cancer exosomes from AD-MSCs block the cell cycle and activate apoptotic signalling. In turn, exosomes derived from ovarian cancer cells induce both the phenotypic and functional transformation of AD-MSCs into a tumour-associated myofibroblastic cell phenotype. Secreted exosomes are represented by asterisks in yellow (adipocytes), blue (ovarian cancer cells) and green (AD-MSCs). Cancer cell–related changes are shown in boxes.

The role of exosomes secreted from adipose mesenchymal stem cells (AD-MSCs) and pre-adipocytes in regulating tumour cell behaviour has been addressed [50,51]. Injection of exosomes from 3T3-L1 pre-adipocytes into the mammary fat pad together with MCF10-DCIS breast cancer cells in vivo favours primary tumour growth [50]. Treatment of MCF-7 breast cancer cells with AD-MSC exosomes in vitro enhances cell migration in a dose-dependent manner, inducing signalling pathways associated with tumour progression like the Wnt/β-catenin pathway (figure 2b) [51]. Hence, exosomes secreted by AD-MSCs have a notable influence on breast cancer tumour cells, reinforcing their migration and growth. Nevertheless, the concentration of exosomes used in some of these studies is higher than the standard doses used elsewhere, making some of the results difficult to interpret. Thus, it is crucial to determine what doses of exosomes are representative of the physiological situation generated by their secretion from MSCs in vivo.

Alternatively, AD-MSC exosomes reduce the viability and proliferation of A2780 and SKOV-3 ovarian cancer cells, as well as their wound-repair and colony-forming abilities [52]. In this model, exosomal miRNAs are crucial regulators of cell-cycle progression and cancer cell survival, promoting anti-tumour effects, highlighting the controversial role of MSCs in tumour development. AD-MSCs exosomes can also induce the upregulation of several pro-apoptotic signalling molecules (BCL2-associated X (BAX), caspase 9 (CASP9), and CASP3) and the downregulation of the anti-apoptotic molecule BCL2, thereby activating apoptotic signalling (figure 2c, orange arrow) [52]. Tumour-derived exosomes also influence the behaviour of AD-MSCs. In ovarian cancer, exosomes derived from 2 different ovarian cancer cell lines induce both the phenotypic and functional transformation of AD-MSCs into a tumour-associated myofibroblastic cell phenotype (figure 2c, blue arrow) [53].

Recently, the exosomes secreted by adipose tissue in physiological conditions somehow exhibited specific organotropism [54]. In this study, miRNAs in adipose tissue–derived exosomes regulate gene expression in specific distant organs, producing a novel means of intercellular communication by EVs similar to mechanisms mediated by adipokines [54]. It would be interesting to determine if the expression of specific molecules on the surface of adipose-derived exosomes (e.g. integrins) could be involved in this organotropic homing, as it was recently described that tumour-derived exosomes drive organ-specific metastasis during pre-metastatic niche formation [10].

Similarly, visceral adipocytes secrete exosomes with specific miRNAs involved in the regulation of the TGF-β and Wnt/β-catenin pathways [55]. Adipocyte-derived exosomes deregulate the TGF-β pathway in hepatic cells in models of obesity, inducing the expression of genes involved in fibrosis [56]. Interestingly, TGF-β induces pro-inflammatory signalling and fibrosis during pre-metastatic niche formation in the liver [12]. To date, it is unclear whether adipocyte-derived exosomes stimulate pre-metastatic niche formation in the liver in conditions of obesity. Nevertheless, evidence suggests that EVs secreted by adipose tissue have a paracrine effect that modulates distant cells and tissues, contributing to tumour progression, and that obesity may reinforce this communication.

4. Understanding the effect of tumour cells in adipose tissue

Adipocytes represent a significant part of the tissue surrounding a tumour; however, their role in supporting tumour progression and metastasis by providing metabolic substrates is poorly understood. Such process is particularly relevant in breast cancer, where tumour cells and adipocytes undertake reciprocal metabolic crosstalk [57]. Thus, breast cancer tumour cells induce changes in the surrounding adipocytes, including delipidation and conversion towards a cancer-associated adipocyte (CAA) phenotype [42,57]. These cells are characterized by a lower lipid content, fewer late adipose markers, and overexpression of inflammatory cytokines and proteases (figure 3) [42]. In turn, mature adipocytes secrete free fatty acids (FAs) that are transferred to tumour cells (figure 3), stimulating FA metabolism through the enhanced expression of carnitine palmitoyltransferase 1A (CPT1A) and proteins of the electron transport chain (ETC) complex, as well as promoting tumour cell proliferation and migration in vitro [57].

Figure 3.

Breast cancer tumour cells induce adipose tissue delipidation. Tumour cells induce phenotypic changes in the surrounding adipocytes, including delipidation and conversion towards cancer-associated adipocytes (CAA). In turn, mature adipocytes secrete free fatty acids, stimulating fatty acid metabolism and the upregulation of chemoattractants in tumour cells, increasing their malignant potential (see text for more details). Cancer cell–related changes are shown in purple boxes. Changes related to adipose tissue are shown in yellow boxes.

CAAs also stimulate the invasiveness of tumour cells in vitro and they enhance their metastatic potential in breast cancer models in vivo [42]. The continued crosstalk between tumour cells and CAAs generates fibroblast-like cells (referred to as adipocyte-derived fibroblasts, ADFs) through the activation of the Wnt/β-catenin pathway in response to Wnt3a secreted by the tumour (figure 3, upper panel detail). ADFs overexpress type I collagen and fibronectin, and they progressively increase their migratory and invasive capacity [58]. Importantly, it was proposed that ADFs can migrate to the middle of the tumour and collaborate in the desmoplastic reaction [58]. Similarly, the in vitro co-culture of ovarian cancer cells with adipocytes induces lipolysis and the transfer of lipid droplets to tumour cells, fuelling tumour growth [27]. The transfer of FAs induces β-oxidation and stimulates the upregulation of FABP4 in omental metastases related to primary ovarian tumours, suggesting that FABP4 fulfils a key role in ovarian cancer metastasis [27].

Interestingly, exosomes secreted by tumours are also involved in the lipolytic processes occurring in adipocytes. Specifically, pancreatic cancer cells induced lipolysis in subcutaneous adipocytes through a mechanism involving exosomal adrenomedullin [59]. Likewise, lung cancer–derived exosomes inhibit adipogenesis of human AD-MSC through a TGFβ-dependent mechanism, which defines a new process by which tumour exosomes can induce changes in adipose tissue [60]. However, whether this process favours tumour progression remains to be determined. Thus, tumour cells can modulate the characteristics of surrounding adipose tissue to a tumour-supportive phenotype. Tumour cells seem to promote lipolysis in surrounding adipocytes, providing FAs that fuel rapid tumour growth. Identifying the main molecules involved in tumour-associated lipid metabolism and transport would give us new clues and targets for the treatment of cancers where obesity could be a major microenvironmental factor.

5. Linking obesity and metastasis-initiating cells

Several studies have suggested that the stem cell behaviour of certain tumour cells could be reinforced, at least partially, by adipose tissue stem cells (ASCs) [61,62]. Indeed, the stem-like behaviour and markers in tumour cells could be influenced by immature adipocytes surrounding the tumour [62]. The accumulation of body fat in obese individuals alters the physiological function of the adipose tissue, modifying the biology of adipose stromal cells and ASCs, and affecting adipokine secretion [63]. Thus, many of the cytokines associated with a pro-inflammatory state not only are upregulated in obese adipose tissue but also may stimulate the self-renewal of cancer stem cells (CSCs) [29].

A direct link between metastasis-initiating cells (MICs) and fat intake has been described in oral cancer, where CD36 is overexpressed in a subpopulation of MICs increasing lipid metabolism and metastatic behaviour [64]. CD36 is a membrane receptor whose expression can be modulated by the tumour microenvironment and it is upregulated by obesity, which in turn, enhances FA uptake in tumour cells [65]. Furthermore, CD36 and FA uptake have been associated with tumour progression and the EMT in hepatocellular carcinoma [66]. CSCs are also enriched in CD36, which participates in the signalling that drives glioblastoma self-renewal and tumour initiation through the uptake of oxidized lipoproteins [67]. Importantly, CD36 expression is associated with poor survival and high metastatic potential in patients with different tumour types, revealing the importance of lipid metabolism in tumour progression [64]. The use of a HFD or palmitic acid augments metastasis, as reflected by the increased invasion of CD36⁺ MICs in oral cancer models. Hence, FA intake and oxidation via CD36 in such tumour cells may give them a metabolic adaptive advantage for survival in distant organs. Interestingly, inhibiting CD36 impairs the metastasis of human melanoma and breast cancer–derived tumours, suggesting that therapies targeting lipid metabolism and transporters may block MICs [64]. Recently, it was reported that palmitic acid-treated hepatocytes display increased CD36 expression and exosome production. Exosomes derived from palmitic acid-treated hepatocytes enhanced the expression of fibrosis markers in hepatic stellate cells in a mechanism depending on miRNA 192 [68]. Moreover, in a brief review article including unpublished data, CD36 has been suggested to control exosome uptake in leukaemia cells [69]. These data suggest that the analysis of exosome uptake and regulation of MICs by this scavenger receptor could be a potential mechanism to explore.

6. Obesity in cancer incidence and progression

The concern about obesity has become more general because it has been recognized as the most relevant risk factor for the development of cancer after smoking [70]. In addition to the association with a greater incidence of cancer, being overweight and obesity increase the risk of death in breast, oesophagus, colon, rectum, liver, gallbladder, pancreas and kidney cancer, as well as non-Hodgkin lymphoma and multiple myeloma [23]. Besides the increased risk of developing such cancers, obesity complicates the treatment of cancer patients, making it more difficult to determine the correct doses for chemotherapy [71]. First-line metastatic chemotherapy achieves better results in non-obese versus obese patients, in terms of treatment response and survival [72]. Further studies should analyse larger samples to evaluate the associations between key tumour characteristics and obesity, as well as consider patterns of weight change over the life course.

Being overweight is, therefore, understood as a carcinogenic factor in many tumours, and whether it precludes or influences the progression to metastasis is yet to be established. The few studies evaluating the relationship between obesity and mortality (and mortality due to metastasis) differ in their accuracy, methodology and conclusions. However, such an association has been described in pancreatic cancer patients [73] and triple negative and HER2-positive breast cancer patients [72,74,75]. Specifically, in breast cancer, obesity constitutes an independent factor of adverse prognosis in high-risk patients [75]. Obesity has been linked with visceral metastasis, and a worse response to chemotherapy and survival [72,75]. For instance, data from 60 000 patients with ER⁺ metastatic disease included in 70 different clinical trials show that obese women have a 34% higher risk of dying from breast cancer [76].

Although research indicates there is only a weak positive association between adiposity and the risk of ovarian cancer, emerging data suggest that only a higher BMI may be associated with a risk of certain histological subtypes, including low-grade serous and invasive mucinous tumours [77]. Interestingly, some larger studies and meta-analyses have reported a stronger relationship with premenopausal ovarian cancers, which are more likely to be of these subtypes [78,79]. The World Cancer Research Fund concluded in its 2014 update that a probable causal relationship exists between body fatness measured by BMI and ovarian cancer, and that this effect may vary in the function of histology, hormonal replacement therapy and menopausal status. Furthermore, being overweight was correlated with shorter progression-free survival (PFS) and OS in platino-sensitive ovarian cancer patients [45]. Indeed, circulating leptin levels were positively correlated with ascites levels, which could explain the systemic and local effects of obesity [45].

Although there is an association between obesity and melanoma progression, clinical studies into the impact of obesity in melanoma patients have been scarce and contradictory. A higher percentage of melanoma was found in obese men [80], although not statistically significant, there is an inverse association between melanoma incidence and serum levels of adiponectin, which has been proposed to have antineoplastic effects [80]. While the current first-line treatment for stage IV melanoma happens to be immunotherapy, bevacizumab (an anti-angiogenic agent) plus interferon was routinely used in 2015, and there was a statistically significant association between an increased VFA/SFA (visceral fat area/subcutaneous fat area) ratio and a worse PFS (p = 0.009) and OS (p = 0.007) associated with that treatment. Notably, a prognostic score, combining VFA/SFA, lactate dehydrogenase (LDH) levels and the presence or absence of liver metastases, was very accurate in predicting PFS at 3 months (AUC 0.759) and OS at 24 months (area under the curve (AUC) 0.846), more than LDH and liver metastases alone (PFS, AUC 0.705; OS, AUC 0.786) [81]. Hence, it is necessary to further investigate the impact of obesity-related factors in the progression of melanoma. Unfortunately, none of the clinical studies performed so far have analysed EVs. It would be of interest to include a standardized analysis of circulating EVs in the clinical setting, and to set up biobanks of plasma and serum to prospectively or retrospectively analyse the relevance or utility of EVs in this setting.

7. Final remarks

Links between increased adiposity and cancer progression have not yet been fully defined. However, there is sufficient clinical and preclinical evidence to encourage further research into this issue, especially because excess weight is increasingly affecting the population and it is a modifiable factor of risk. On the one hand, it will be important to study what problems must be surpassed in epidemiological studies, such as better anthropometric data collection on body fatness in CT or MRI scans and visceral-abdominal fat versus BMI (as an indirect measure), the timing of obesity (childhood, young adulthood etc.) and the cumulative effect of obesity over the course of a lifetime. On the other hand, translational studies that help to define the underlying molecular mechanisms of adiposity in cancer are vital in the field.

It was recently proposed that EVs from adipose tissue in non-cancer environments have a specific organotropic behaviour [54]. It would be interesting to assess if there are specific signals delivered from adipose tissue in a tumour setting and if organs affected by such adipose-derived signals (e.g. adipokines, EVs) could trigger pre-metastatic niche formation. One of the main limitations in these studies is that although preclinical data suggest that adipose tissue secretes factors and vesicles during metastasis, linking obesity with metastasis in a clinical setting is often hard to establish. Determining the relevance of these mechanisms in a clinical setting would be crucial to define the specific molecules involved. Surprisingly, most of the preclinical studies have been performed in vitro, and with doses that do not clearly reflect EV physiological levels. Therefore, interpreting the impact of such molecules on metastasis and in in vivo models is difficult; defining the role of these molecules in animal models of metastasis should help to define their role in human metastasis.

Interestingly, data suggest that MICs are a subpopulation of cells that rely specifically on lipid metabolism to survive. Determining which molecules (e.g. CD36) are crucial for their viability and maintenance could be important to find compounds that destroy cancer progenitor cells, thereby preventing metastasis. Mechanistically, MICs seem to depend on lipid metabolism to obtain more energy, which may favour their survival at sites distant from the primary tumour [64]. Whether vesicles secreted by adipose tissue could reinforce lipid metabolism in tumour cells and metastatic behaviour is a plausible hypothesis that needs to be tested.

In summary, we can identify at least three mechanisms involved in the crosstalk between tumour and adipose tissue during metastasis: (i) The influence of pro-inflammatory molecules on the metastatic behaviour of the primary tumour; (ii) the enhanced communication and exchange of molecules via EVs; and (iii) Adipose tissue delipidation and tumour cell fuelling. However, it remains unclear whether obesity has a direct impact on metastatic behaviour or if it is an additional environmental factor. Further research will be needed to define the molecular mechanisms involved in this relationship and to identify potential therapies that could revert metastatic cell behaviour.

Data accessibility

This article has no additional data.

Competing interests

We declare we have no competing interests.

Funding

The authors gratefully acknowledge support from the following funding sources: the US National Cancer Institute (CA169416), MINECO (SAF2014-54541-R), ATRES-MEDIA – AXA, Asociación Española Contra el Cáncer, WHRI Academy, Worldwide Cancer Research and Fundación de Investigación Oncológica FERO.