Background
Cutaneous melanoma is a highly aggressive disease of increasing incidence caused by malignant transformation of epidermal melanocytes [
1,
2]. The disease progresses through distinct steps, and the transition from the radial growth phase to the vertical growth phase is associated with tumor-induced angiogenesis, increased metastatic propensity, and significant worsening of the prognosis [
3,
4]. Primary melanomas in the vertical growth phase secret angiogenic factors that stimulate hem- and lymphangiogenesis, resulting in the development of a dense intratumoral network of blood vessels and an enriched peritumoral network of lymphatics [
5‐
7]. High rates of hem- and lymphangiogenesis have been shown to be associated with metastatic growth and poor survival [
8‐
13]. Recent reviews have concluded that the blood vessel density (BVD) in the tumor periphery and the peritumoral lymph vessel density (LVD) have significant prognostic power in patients with melanoma, whereas the BVD in central tumor regions and the intratumoral LVD may not have prognostic value [
14‐
16].
Intravital microscopy studies have revealed that the blood vessel networks of melanomas in human patients show severe morphological abnormalities, including vessel disorganization, aberrant vessel bifurcations, heterogeneous vessel density, vessel tortuosity, increased vessel segment lengths, and highly permeable vessel walls [
17]. These abnormalities cause irregular and heterogeneous blood flow, and may lead to the development of a physicochemical tumor microenvironment characterized by poor oxygenation and elevated interstitial fluid pressure (IFP) [
18,
19]. Indeed, clinical investigations have revealed that low oxygen tension, hypoxic tumor regions, and high IFP are characteristic features of the physicochemical microenvironment of human melanomas [
20‐
23].
Human melanoma xenografts are frequently being used as preclinical models of melanomas in patients, and it has been shown that the blood vessel and lymphatic networks of melanomas transplanted orthotopically to athymic nude mice are similar to those reported for human melanomas [
24‐
26]. It has also been shown that orthotopic melanoma xenografts develop hypoxic regions and elevated IFP [
27,
28], and furthermore, that metastatic dissemination and growth is associated with high angiogenic activity, extensive hypoxia, and high IFP in the primary tumor [
29‐
31]. These findings have led to the suggestion that there is a link between the angiogenic signature, vascular morphology and function, and the physicochemical microenvironment in melanomas, and that interactions between these biological features may lead to a hostile tumor microenvironment promoting metastasis to regional lymph nodes and distant organs [
32].
Cutaneous melanoma disseminates through blood vessels and lymphatics, and can show an organ specific metastatic pattern [
33]. The time course to the development of distant metastases may differ between the metastatic pathways [
34], and the progression of melanomas spreading by the hematogenous route is believed to be fast compared with that of melanomas spreading by the lymphogenous route [
33‐
35]. Several genetic biomarkers associated with the metastatic route and organ specific metastasis have been identified [
33]. However, studies examining the possibility that the metastatic pathway of melanomas is influenced significantly by the vascular and physicochemical tumor microenvironments are sparse.
In a previous study, we searched for associations between the overall metastatic propensity, the angiogenic potential, and the hypoxic tumor microenvironment of melanomas by using nine different melanoma xenograft lines as preclinical tumor models [
36]. The study revealed that the metastatic propensity was determined by the tumor microvascular density rather than the fraction of hypoxic tissue, and vascular endothelial growth factor-A and interleukin-8 were identified as important drivers of tumor angiogenesis. In the study reported here, we addressed the possibility that the metastatic pathway and organ specific metastatic pattern of melanomas is influenced by the microvascular and physicochemical tumor microenvironments by searching for associations between metastatic spread and BVD, LVD, IFP, fraction of hypoxic tissue, and angiogenic signature. Orthotopic patient derived xenograft (PDX) models and cell line-derived xenograft (CDX) models of cutaneous melanoma were examined, and we provide significant evidence that tumor angiogenesis and the physicochemical tumor microenvironment may have strong impact on the metastatic pathway of melanomas.
Discussion
Hematogenous and lymphogenous metastatic spread are characteristic features of angiogenic human melanomas [
33,
34], and the possibility that the metastatic route is associated with the microvascular and physicochemical microenvironments of the primary tumor was investigated in this study by using orthotopic melanoma xenografts as preclinical models of human disease. C-10, D-12, and E-13 tumors disseminated primarily by the hematogenous route and developed pulmonary metastases, whereas N-15, R-18, and T-22 tumors mainly showed lymphogenous metastatic spread. These two tumor groups did not differ significantly in HF
Pim, peripheral BVD, or peritumoral LVD, suggesting that tumor hypoxia and microvascular density are not important determinants of the primary metastatic route of melanomas.
Because the peripheral BVD of C-10, D-12, and E-13 tumors was similar to that of N-15, R-18, and T-22 tumors and C-10, D-12, and E-13 tumors grew faster and showed higher expression of angiogenesis-related genes than N-15, R-18, and T-22 tumors, the angiogenic activity was higher in C-10, D-12, and E-13 tumors than in N-15, R-18, and T-22 tumors. Consequently, hematogenous metastatic spread of melanomas may be associated with high angiogenic activity in the primary tumor. This suggestion is in accordance with clinical studies having shown higher angiogenic activity in the primary tumor in melanoma patients with distant organ metastases than in those with lymph node metastases only [
39]. High angiogenic activity may promote hematogenous metastasis of tumors by several mechanisms. First, fast formation of new blood vessels leads to the development of an immature microvasculature with fragmented basement membrane, leaky vessels, and high density of vessel sprouts, morphological characteristics that may facilitate tumor cell intravasation [
40]. Furthermore, elevated capacity to induce neovascularization may increase the probability of tumor cells trapped in secondary organ capillary beds to extravasate and give rise to macroscopic growth [
41].
IFP was substantially higher in N-15, R-18, and T-22 tumors than in C-10, D-12, and R-18 tumors, suggesting that lymphogenous metastatic spread of melanomas is associated with highly elevated IFP in the central regions of the primary tumor. High IFP has been shown to promote lymph node metastasis also in human cervix carcinoma and pancreatic carcinoma xenografts [
42,
43]. Mechanisms linking high IFP in tumors to lymph node metastasis have not been identified conclusively, but several possible mechanisms have been suggested. First, high tumor IFP may force interstitial fluid to flow from the tumor tissue into adjacent normal tissues [
42], and this fluid flow may direct tumor cells toward peritumoral lymphatics by autologous chemotaxis [
44]. Moreover, the interstitial fluid may transport proteolytic enzymes and chemokines that facilitate tumor cell migration by remodeling the extracellular matrix [
44], and may carry lymphangiogenic factors that promote metastasis by dilating peritumoral lymphatics and inducing lymphangiogenesis [
45].
Tumors develop elevated IFP because they show high resistance to blood flow, low resistance to transcapillary fluid flow, and impaired lymphatic drainage, and the intertumor heterogeneity in IFP is primarily a consequence of differences in blood flow resistance [
46]. Low-diameter vessels are the main cause of high resistance to blood flow, but abnormal vessel bifurcations, vessel tortuosity, and long vessel segment lengths may also contribute significantly [
47]. The resistance to blood flow was most likely higher in N-13, R-18, and T-22 tumors than in C-10, D-12, and E-13 tumors, resulting in higher IFP in the former group of tumors. Taken together, our observations suggest that melanomas disseminating primarily by the lymphogenous route develop a microvascular network that exerts high resistance to blood flow and causes highly elevated IFP, whereas melanomas disseminating primarily by the hematogenous route have highly elevated angiogenic activity that results in a microvascular network with dilated vessels that exerts lower resistance to blood flow and facilitates tumor cell intravasation.
The expression of ANGPT2, F3, and TIE1 was more than twofold higher in C-10, D-12, and E-13 tumors than in N-15, R-18, and T-22 tumors, and the expression of NRP2 was more than twofold higher in N-15, R-18, and T-22 tumors than in C-10, D-12, and E-13 tumors. These observations suggest that high expression of F3 and genes of the angiopoietin–tie system is associated with high angiogenic activity, hematogenous metastatic spread, and the development of pulmonary metastases in melanoma xenografts, whereas high expression of NRP2 is associated with lymphogenous metastatic spread and the development of lymph node metastases. Further studies are needed to ascertain whether there is a causal relationship between elevated expression of these genes and the metastatic route of melanomas.
The angiopoietin–tie system plays important roles in vascular development, morphogenesis, and homeostasis, and is implicated in several diseases where the vasculature is important, including cancer [
48,
49]. Blocking of ANGPT2 protein or genetic deletion of TIE1 has been shown to decrease tumor angiogenesis and growth by reducing endothelial cell sprouting and by inducing endothelial cell apoptosis and vessel regression [
50,
51]. It has also been revealed that blocking of ANGPT2 protein may induce vascular normalization in tumors [
52], inhibit transendothelial migration of tumor cells [
53], and decrease pulmonary metastasis and growth of experimental tumors [
54‐
56]. The association between hematogenous metastatic spread of melanomas and high expression of ANGPT2 and TIE1 reported here is consistent with these observations.
F3 encodes coagulation factor III, also known as tissue factor (TF). Full length TF promotes tumor angiogenesis by up-regulating the expression of several proangiogenic factors, including interleukin-8, chemokine (C-X-C motif) ligand-1, and vascular endothelial growth factor-A [
57,
58]. Spliced TF may stimulate tumor angiogenesis through protein tyrosine kinase-2 signaling by ligating to endothelial cell integrins such as αvβ3 and α6β1 [
57,
59]. In addition, TF may play an important role in tumor cell intravasation [
60]. The significance of TF in lung metastasis of melanoma has been studied previously [
61,
62]. TF was found to promote lung colonization of melanoma cells inoculated intravenously into immunodeficient mice, but did not facilitate spontaneous lung metastasis of subcutaneous melanomas. It is thus possible that the association between pulmonary metastasis and high expression of F3 reported here was due to effects of TF in the later phases of the metastatic process after the melanoma cells had entered the blood circulation as well as effects of TF within the primary tumor.
NRP2 encodes the neuropilin-2 protein, which is a transmembrane multifunctional nonkinase receptor for class III semaphorins, members of the vascular endothelial growth factor family, and other growth factors [
63]. Tumor-induced activation of neuropilin-2 on lymphatic endothelial cells can increase peritumoral lymphangiogenesis and promote lymph node metastasis [
64]. Neuropilin-2 on tumor cells interacts with integrins on blood and lymph vessel endothelial cells to mediate vascular adhesion and promote extravasation [
65]. High expression of NRP2 has been shown to promote hematogenous metastatic spread in pancreatic adenocarcinoma and clear cell renal cell carcinoma xenografts [
65] and the development of lymph node metastases in patients with breast carcinoma [
66], papillary thyroid carcinoma [
67], and squamous cell carcinoma of the oesophagus [
68]. Immunohistochemical investigations have revealed that melanomas show high expression of neuropilin-2 and that the expression is higher in lymph node metastases than in the primary tumor [
69]. These observations suggest that the association between high expression of NRP2 and lymph node metastasis reported here was not due to effects of neuropilin-2 within the primary tumor, but most likely was due to interaction between melanoma cells and lymphatic endothelial cells after the melanoma cells had entered the peritumoral lymphatic network.
Authors’ contributions
RH, LMKA, EKR developed the original hypothesis and experimental design. RH, LMKA carried out experiments. RH, EKR analyzed data. RH, EKR wrote the manuscript. All authors read and approved the final manuscript.