Role of lymphatic vasculature in regional and distant metastasis

Introduction

Metastasis is the main cause of treatment failure and death for cancer patients. The involvement of lymphatic system with cancer has long been recognized as an important indicator of cancer aggressiveness. Lymph node status is one of the key parameters used for determining stage of disease progression and it is a powerful predictor of patient survival (Edge, 2010). Patients with lymph node metastases are also more likely to present with disease recurrence (Rosen, 2008). However, the causal link between lymphatic dissemination and the negative outcome is not understood and how exactly the lymphatic system contributes to cancer progression from localized to systemic, disseminated disease remains a critical open question. Although the number of publications on the topic of cancer lymphatics has been growing steadily over the past decade, there is still a lot to be learned. This review highlights advances in our understanding of the mechanisms by which lymphatic vessels, and in particular lymphatic endothelium, impact metastasis.

Tumor lymphangiogenesis

Upon identification of VEGF-C and VEGF-D as lymphangiogenesis factors(Jeltsch et al., 1997; Joukov et al., 1996; Joukov et al., 1997), we and others have reported more than a decade ago that induction of lymphangiogenesis by the tumor facilitates metastatic spread (Mandriota et al., 2001; Skobe et al., 2001; Stacker et al., 2001). Since then, work from many laboratories has recapitulated these findings in numerous animal models and further showed that inhibition of lymphangiogenesis by blockade of VEGF-C or its receptor VEGFR-3, prevents lymph node metastases without significantly affecting primary tumor growth (Brakenhielm et al., 2007; Burton et al., 2008; Chen et al., 2005; He et al., 2005; Kawakami et al., 2005; Krishnan et al., 2003; Lin et al., 2005; Mandriota et al., 2001; Mattila et al., 2002; Skobe et al., 2001; Yanai et al., 2001). VEGF-C also facilitates metastatic spread to distant sites and, conversely, blocking VEGF-C or VEGFR-3 inhibits distant metastases in majority of experimental models (Brakenhielm et al., 2007; Burton et al., 2008; Chen et al., 2005; Krishnan et al., 2003; Lin et al., 2005; Roberts et al., 2006). In agreement with the preclinical data, numerous clinical studies reaffirmed the negative correlation between VEGF-C, lymphangiogenesis and patient outcome (Alitalo and Carmeliet, 2002; Ding et al., 2007; Furudoi et al., 2002; Miyazaki et al., 2008; Mohammed et al., 2007; Pepper et al., 2003; Swartz and Skobe, 2001; Tsutsumi et al., 2005). VEGF-C and VEGF-D are most specific and best studied lymphangiogenesis factors, however, tumor lymphangiogenesis can be mediated also by several pleiotropic factors, including PDGF-BB, IGFs, FGF2, HGF, Ang2, adrenomedulin and IL-7 (Zheng et al., 2014).

Lymphangiogenesis associated with the primary tumor is thought to increase metastasis by increasing the probability for tumor cells to enter into the lymphatic vessels. Large numbers of newly generated lymphatics create more opportunities for tumor cell exit and close proximity of tumor cells to LECs could make more tumor cells respond to LEC-derived chemokines and be mobilized into the lymphatics. Furthermore, gene-profiling data of tumor-activated and quiescent lymphatic endothelium showed significantly different expression profile, suggesting that tumor cells may interact differently with the pre-existing and with the newly formed lymphatics (Clasper et al., 2008). The nature and significance of that cross-talk, however, remain to be elucidated. Importantly, while tumor lymphangiogenesis profoundly increases metastatic spread, it is not an obligatory step for metastasis. Controversy on this topic stems from the assumption that if angiogenesis is required for tumor growth, by inference, lymphangiogenesis must be a requirement for metastasis. However, paradigms established for tumor angiogenesis cannot be extrapolated on lymphangiogenesis, since function of lymphatics and blood vessels in tumors is very different despite the fact that the endothelial biology of these two vascular systems is shared on many levels.

Interestingly, lymphangiogenesis in the sentinel lymph nodes has been shown to precede lymph node metastasis in several studies(Dadras et al., 2005; Harrell et al., 2007; Hirakawa et al., 2007; Hirakawa et al., 2005; Ruddell et al., 2008; Van den Eynden et al., 2006; Van den Eynden et al., 2007). Lymph node lymphangiogenesis is a component of the normal host immune response (Angeli et al., 2006; Kim et al., 2012; Randolph et al., 2005), which in the tumor setting is thought to enhance metastasis by creating a pre-metastatic niche. Because selective inhibition of lymph node lymphangiogenesis is difficult to achieve, this concept is derived mainly from correlative studies and more work is needed to elucidate exact mechanisms and roles of LN lymphangiogenesis in cancer spread. Lymphangiogenesis has also been documented within metastases in the sentinel and more distal lymph nodes (Kerjaschki et al., 2011). Furthermore, this study indicated that tumor cell invasion into the intrametastatic lymphatic vessels and formation of tumor emboli is necessary for metastatic dissemination into more distal lymph nodes (Kerjaschki et al., 2011).

Mechanisms of lymph node metastasis

Many important questions about lymph node metastasis remain unresolved to date. Lymph nodes are usually first sites of detectable metastases, which could be due to the preference of tumor cells to enter into the lymphatic vessels. It is not known however, whether such preference exists and whether tumor cell rate of entry into the lymphatic and blood vessels is different. Alternatively, early metastasis in the lymph nodes could be a result of survival or growth advantage within the lymph node microenvironment. Another key unresolved question is to which extent lymph node metastases directly contribute to the formation of distant metastases. While these issues have been frequently debated, there is no data to clearly support or oppose any of the aforementioned concepts.

Over decades, lymphatics were portrayed as passive participants in metastasis and regarded mainly as a transportation system. Recent studies, however, indicate that tumor cells are guided into the lymphatic vessels by chemokines produced by lymphatic endothelium (Ben-Baruch, 2008; Das and Skobe, 2008). CCL21 is constitutively expressed by the lymphatic vessels (Gunn et al., 1998; Kerjaschki et al., 2004; Podgrabinska et al., 2002; Shields et al., 2007), immobilized by binding to heparin sulfates and forms steep gradients within the perilymphatic interstitium (Haessler et al., 2011; Schumann et al., 2010; Weber et al., 2013). These gradients induce directed migration of dendritic cells towards lymphatics from a distance of up to 90 microns (Weber et al., 2013), suggesting that melanoma and breast cancer cells expressing CCR7 receptor (Houshmand and Zlotnik, 2003; Muller et al., 2001) could also be guided into the vessels by such haptotactic chemokine gradients. Overexpression of CCR7 in melanoma has indeed been shown to facilitate lymph node metastasis in a mouse model (Wiley et al., 2001) and clinical studies have confirmed the correlation between CCR7 expression and lymph node metastasis (Cabioglu et al., 2005; Ishigami et al., 2007; Mashino et al., 2002). CXCL12 is another chemokine that has been shown to facilitate lymph node metastasis of CXCR4+ tumor cells (Hirakawa et al., 2009; Muller et al., 2001; Uchida et al., 2007). CXCL12 is upregulated on lymphatic vessels in the primary tumor and it has been shown to promote recruitment of CXCR4+/CD133+ melanoma cells into the proximity of lymphatic endothelium. However, direct evidence for its role in directing cells into the lymphatic capillaries is lacking. Several studies suggested that macrophage mannose receptor I (MR) and CLEVER-1 may be important mediators of cancer cell adhesion to lymphatic endothelium (Irjala et al., 2003; Irjala et al., 2001). MR and CLEVER-1 expression has been detected on tumor lymphatic vessels and it was associated with increased lymph node metastases (Irjala et al., 2003). There is no evidence, however, that adhesive interactions with LECs are indeed required for tumor cell entry into the lymphatics and the mechanisms of tumor cells intravasation into the lymphatic vessels remain elusive. Conventional wisdom implies that tumor cells will be delivered into the sentinel lymph nodes with the flow of lymph once they are inside the lymphatic lumen, and this has indeed been demonstrated for tumor cell transport within large, collecting lymphatic vessels (Dadiani et al., 2006; Hayashi et al., 2007). In lymphatic capillaries, however, dendritic cells have been shown to crawl along the luminal side of LECs towards lymph node in the direction of flow (Pflicke and Sixt, 2009; Tal et al., 2011), opening the possibility that tumor cells could employ similar mechanisms.

Subcapsular sinus (SCS) of the LN, which is lined by LECs, is the first site of lymph node metastasis (Carr, 1983; Carr et al., 1985; Dadiani et al., 2006; Das et al., 2013; Dewar et al., 2004). Dilation of SCS, which starts at the junction with the afferent lymphatic vessel, precedes arrival of tumor cells (Das et al., 2013) and may be a prerequisite for allowing the entry of tumor cells into the SCS. Indeed, in the absence of the primary tumor, when injected directly into the lymphatic system, osteosarcoma and melanoma cells arrest at the junction of the afferent lymphatic vessel and the LN (Hayashi et al., 2007). Scanning Electron Microscopy (SEM) analysis revealed that SCS is divided vertically and horizontally into smaller compartments, resulting in passages 5–15 micron wide (Das et al., 2013; Jia et al., 2012; Ohtani and Ohtani, 2008). Since the diameter of a single circulating tumor cell is at least 15 micron (Vona et al., 2000), it has been concluded that the small dimensions of the sinus prevent passive flow of tumor emboli into the SCS (Das et al., 2013). Chemokine CCL1 produced by the SCS LECs facilitates tumor cell entry into the open SCS as well as subsequent migration across the floor of the sinus into the LN cortex. Conversely, blocking CCR8, which is expressed in a large subset of melanomas, led to the arrest of tumor cells at the junction of the afferent lymphatic vessel and the LN (Das et al., 2013). These studies demonstrate that LECs of the LN SCS represent a barrier for entry of tumor cells into the lymph node and identify novel function for CCL1-CCR8 in controlling the egress of tumor cells from the afferent lymphatics (Figure 1). From the sentinel lymph node, metastatic cells advance into the subsequent LNs by the mechanisms that are not well understood. Chemokine CXCL12 could play a role in LN exit since it is upregulated on LN lymphatics when metastases are present (Kim et al., 2010). Ductal mammary carcinomas were shown to invade in bulk into the LN lymphatics through large openings in the lymphatic vessel’s walls (Kerjaschki et al., 2011; Yamaguchi et al., 2005). Large gaps in LECs were induced by the tumor-derived arachidonic acid metabolite 12S-HETE (Kerjaschki et al., 2011).

Figure 1. Model for tumor cell entry into the lymph node.

Prior to the arrival of tumor cells, subcapsular sinus (SCS) dilates starting at the orifice of the afferent lymphatic vessel. Tumor emboli arriving from the afferent lymph first arrest at the junction of the afferent lymphatic vessel and the subcapsular sinus. From here, tumor cells expressing CCR8 migrate laterally into the subcapsular sinuses, guided by the CCL1 chemokine which is presented on the surface of SCS LECs. Single cells and small cell clusters may move with the flow of lymph laterally into the sinus. Within the SCS, tumor cells attach to the floor and the roof of the sinus, where they continue to proliferate. Colonization of the SCS is a first step of lymph node metastasis and it is a result of concurrent migration and growth of tumor cells within the sinus. Next step is tumor cell migration across the floor of the sinus into the LN cortex, process also guided by the CCL1 chemokine presented by SCS LECs.

Lymphatics and distant metastasis

From the therapeutic perspective, it is critically important to understand what role lymphatics play in the formation and progression of distant metastases. Thus far, experimental evidence demonstrated that in most cases inhibiting lymph node metastases diminished incidence of lung metastases (Brakenhielm et al., 2007; Burton et al., 2008; Chen et al., 2005; Krishnan et al., 2003; Lin et al., 2005; Roberts et al., 2006), indicating that inhibiting lymphatic dissemination could be a promising approach for preventing distant metastases in certain patients. Concurrent inhibition of lymphangiogenesis and angiogenesis by inhibition of VEGFR-3 and VEGFR-2, respectively, has been shown to effectively diminish lung metastases in the intervention treatment regimen (Matsui et al., 2008; Roberts et al., 2006), suggesting that such combination approach could attenuate metastatic disease also in certain patients with established lung metastases. While it is clear that lymphatics contribute to the early stages of metastases serving as a route for dissemination from the primary tumor to the regional lymph nodes and possibly for the subsequent spread to distant sites, it is less well understood what role lymphatics in distant organs play for already disseminated disease.

In some patients, metastatic disease in the lung is characterized by extensive involvement of lung lymphatics with cancer (Acikgoz et al., 2006; Bruce et al., 1996; Goldsmith et al., 1967; Janower and Blennerhassett, 1971; Thomas and Lenox, 2008; Tomashefski and Dail, 2008). This type of metastasis is referred to as pulmonary lymphangitic carcinomatosis and it is most commonly observed in patients with breast, lung, gastric, pancreatic and prostate cancer (Thurlbeck, 1979; Tomashefski and Dail, 2008). Strikingly, most of these patients die within several months of diagnosis (Bruce et al., 1996; Thomas and Lenox, 2008; Tomashefski and Dail, 2008; Yang and Lin, 1972). How frequent this type of metastasis is in the patient population, however, is a subject of a debate. Studies reported the incidence of lymphangitic spread to be as low as 6%(Harold, 1952; Minor, 1950; Yang and Lin, 1972) and as high as 56%(Fichera and Hagerstrand, 1965). Because a hallmark of lymphangitic spread is its diffuse presentation, it is very difficult to diagnose in patients with current imaging techniques. For example, 50% of the cases of histologically proven pulmonary lymphangitic carcinomatosis present with normal radiographs (Amundson and Weiss, 1991; Fichera and Hagerstrand, 1965; Goldsmith et al., 1967; Janower and Blennerhassett, 1971; Thurlbeck, 1979; Trapnell, 1964). Because of these imaging limitations in patients and because histologic sampling of lung metastases even at autopsy, is not frequently performed, it is believed that the true incidence of lymphangitic spread is greatly underestimated (Tomashefski and Dail, 2008). Nevertheless, the evidence of lymphangitic carcinomatosis in a patient is invariably associated with extremely poor prognosis, indicating that pulmonary lymphatic vasculature facilitates rapid progression of metastatic disease.

Data from the spontaneous metastasis model in mouse, revealed that overexpression of VEGF-C in tumor cells induced lymphangiogenesis in the lung and changed pattern of metastases to pulmonary lymphangitic carcinomatosis (Das et al., 2010). Expansion of the pulmonary lymphatic network was accompanied with a dramatic increase in size of metastases, which were growing within the constraint of lymphatic vessel walls in the absence of angiogenesis. Together with the clinical observations, these experimental data demonstrate an unappreciated role of lymphatics in facilitating lung colonization. These data also suggest that lymphatic vasculature could be a niche which promotes survival and growth of metastases. Importantly, this opens the possibility that targeting lymphatics could be employed as a strategy for treatment of patients which already have disseminated disease, and not only for prevention of metastatic spread. One study demonstrated an association between CD133+ chemoresistant tumor cells and lymphatics at different metastatic sites (Kim et al., 2010), raising another intriguing possibility that lymphatics may modulate therapeutic response.

Immunoregulatory role of LECs in cancer

Several important functions have been attributed to LECs in the recent years which could influence cancer progression as well as directly impact immunotherapy approaches for cancer. LECs have emerged as important players in directing immune cell traffic from tissues into the lymphatic capillaries (Girard et al., 2012; Johnson and Jackson, 2008; Martin-Fontecha et al., 2009). Best studied chemokine made by LECs is CCL21, which binds to CCR7 on migratory DCs, certain macrophage subsets and T-cells and facilitates directed migration of these cells (Forster et al., 2008; Luther et al., 2000; Nagira et al., 1997; Saeki et al., 1999; Willimann et al., 1998). Importance of CCL21-CCR7 interaction in immunity is illustrated by the fact that mice lacking CCR7 ligands have drastically impaired DC and T cell homing to LNs and cannot mount adaptive immune responses (Forster et al., 2008). LECs express many other chemokines which can attract cells into the lymphatic capillaries (Card et al., 2014), but their exact role in controlling leukocyte traffic is yet to be determined.

In the lymph node, LECs lining subcapsular sinuses direct CCR8+ cells into the LN cortex by presenting CCL1 chemokine to the cells arriving from the afferent lymph (Das et al., 2013). CCR8 has been shown to be important for migration of DCs from the skin into the lymph node (Qu et al., 2004), and since its ligand CCL1 is not made by peripheral lymphatics, it has been concluded that CCL1 made by LECs of SCS controls DC entry into the LN (Das et al., 2013; Jakubzick et al., 2006; Qu et al., 2004). Further evidence for the role of lymph node LECs in guiding and selecting cells for entry into the LN comes from the studies which showed that entry of DCs into the LN occurs preferably across the LECs of afferent SCS, whereas T-cells arriving from afferent lymph preferentially enter via medullary sinuses (Braun et al., 2011). Underlying mechanisms governing this pattern of migration remain to be determined.

LECs of medullary sinuses regulate egress of T-cells from the LN by sphingosine-1-phosphate (S1P). Downregulation of S1P receptor 1 (S1P1) on antigen-activated naïve T cells promotes retention of differentiating T cells in the LN, whereas its upregulation makes cells responsive to the ligand and triggers migration into the cortical sinuses (Schwab and Cyster, 2007) where fluid flow promotes movement of T cells into efferent lymphatic vessels (Grigorova et al., 2009).

In addition to regulating cell traffic, growing body of evidence shows that LECs can directly modulate activity of immune cells and promote tolerance (Girard et al., 2012; Lukacs-Kornek et al., 2011; Norder et al., 2012; Podgrabinska et al., 2009). As DCs enter into and crawl along the lymphatic capillaries, they directly interact with LECs. Under inflammatory conditions, binding of Mac-1 on DCs to ICAM-1 on LECs leads to inhibition of DC maturation and suppresses the ability of DCs to activate T cells (Podgrabinska et al., 2009). Another study demonstrated that supernatant from IFNγ-activated LECs also impaired the ability of DCs to induce allogeneic CD4⁺ T cell proliferation (Lukacs-Kornek et al., 2011). Thus, LECs can regulate T-cell responses by limiting expansion of T-cells in the LNs (Lukacs-Kornek et al., 2011; Norder et al., 2012; Podgrabinska et al., 2009).

LECs express MHC class I (Cohen et al., 2010; Lund et al., 2012; Nichols et al., 2007) and II (Malhotra et al., 2012; Tewalt et al., 2012a) molecules, and can directly induce T cell tolerance as well as suppress T cell activation by expressing several immunoregulatory factors. For example, activated LECs secrete TGF-β, indoleamine-2,3-dioxygenase (IDO) and nitric oxide (NO), all of which are strongly immunosuppressive (Lukacs-Kornek et al., 2011; Malhotra et al., 2012; Norder et al., 2012; Podgrabinska et al., 2002). Furthermore, LECs can modulate T-cell function and induce peripheral tolerance through direct presentation of antigens to T-cells (Cohen et al., 2010; Gardner et al., 2008; Lee et al., 2007; Magnusson et al., 2008; Nichols et al., 2007). In a mouse model, LECs expressed melanoma antigen, tyrosinase epitope Tyr₃₆₉, and induced tolerance of Tyr₃₆₉-specific CD8⁺ T cells (Cohen et al., 2010; Fletcher et al., 2010). Since tyrosinase epitope is a major target for melanoma immunotherapy, these findings suggested that LEC-induced tolerance could have direct impact on the clinical efficacy of anti-melanoma immunotherapies. Interestingly, VEGF-C was shown to promote immune tolerance in B16 F10 murine melanoma which expressed OVA as a foreign antigen; VEGF-C promoted local deletion of OVA-specific CD8(+) T cells and protected tumors against pre-existing antitumor immunity (Lund et al., 2012).

LECs induce tolerance due to high levels of expression of the inhibitory ligand PD-L1 and absence of co-stimulatory molecules on the surface of LECs (Malhotra et al., 2012; Norder et al., 2012; Tewalt et al., 2012a; Tewalt et al., 2012b). Lack of co-stimulation leads to upregulation of programmed cell death 1 (PD-1) receptor expression on CD8 T cells and ultimately antigen-specific deletion of CD8⁺ T cells (Tewalt et al., 2012a). PD-1 is an important negative regulator of T-cell function and a marker of T-cell exhaustion associated with immunosuppression in cancer. PD-1 has emerged as an important target in immunotherapy: inhibition of PD-1 and PD-L1 potently increases anti-tumor CD8+ T-cells effector response and has shown very promising results in cancer patients (Brahmer et al., 2012; Topalian et al., 2012). In addition to PD-L1, LECs were most recently shown to upregulate CTLA-4 in T-cells, another important marker of T-cell exhaustion and a key target for cancer immunotherapy. Together, these findings that multiple inhibitory receptors are expressed at high levels by LECs (Tewalt et al., 2012a) point to an important role of LECs in cancer immunosuppression and indicate that further insight into the mechanisms by which LECs mediate T-cell exhaustion offers potential for discovery of novel therapeutic targets for cancer immunotherapy.

Concluding remarks

Slowly but steadily, perspectives on the role of lymphatics in cancer have been changing. Traditionally viewed only as a transportation system, it has now become clear that lymphatics perform many functions which could profoundly affect cancer progression. Recent discoveries that LECs can modulate adaptive immune responses put lymphatics in the spotlight as a new player in cancer immunoediting. Proper functioning of LECs in controlling immune cell traffic could be important for immunosurveillance during early stages of tumor development and promote host protection against cancer. On the contrary, the ability of LECs to promote immunosuppression could facilitate immune escape and therefore promote tumor initiation, progression and dissemination. In the dynamic interplay with tumor cells and immune cells, LECs may help orchestrate protection against cancer and tumor elimination, but they may also be exploited to facilitate tumor progression. An example for this is lymphangiogenesis, which is anti-inflammatory and an intrinsic component of the immune response, yet in the cancer setting lymphangiogenesis potently augments metastatic spread. Based on the preclinical data showing that blockade of VEGF-C or VEGFR-3 inhibits lymphangiogenesis and metastasis, two humanized blocking antibodies have entered clinical trials in the past year. In view of the more recent data showing that lymphangiogenesis and VEGF-C not only increase regional spread, but also facilitate late steps of metastasis and immunosuppression, it is reasonable to assume that blocking VEGF-C and its receptors could benefit patients with early as well as late stages of cancer as an adjuvant therapy. Importantly, careful design of preclinical studies and clinical trials will be essential for evaluating the potential of targeting lymphatics in cancer. Testing these inhibitors in a broad patient population without attempting to define a subset of patients which are most likely to respond to this type of therapy will inevitably lead to failure. Since the therapeutic potential of targeting lymphatics is just beginning to be explored, we need to maximize the odds for seeing a positive response by assuring that such therapeutics are tested in the most relevant setting.

Abbreviations

CCL

CC Chemokine ligand

CCR

CC Chemokine receptor

CXCR

CXC chemokine receptor

DC

Dendritic cell

IL

Interleukin

LEC

Lymphatic endothelial cells

LN

Lymph node

MR

Mannose receptor

SCS

Subcapsular sinus

SEM

Scanning Electron Microscopy

TNF-α

Tumor necrosis factor alpha