New Vessel Formation

It has been known for nearly a century that the vascular system is associated with tumor growth in animals and humans.¹ Powerful insights into the neovascularization of transplanted tumors using transparent window techniques were developed in the 1940s.^2–5 The possibility that tumors produce an “angiogenic” substance was suggested in 1968.⁶ The hypothesis that blocking angiogenesis should block tumor growth and metastasis was proposed shortly thereafter in 1971.⁷ The concept that a tissue acquires angiogenic capacity during neoplastic transformation—and, by extension, that antiangiogenesis could be used to prevent cancer—was put forward in 1978.⁸ The first antiangiogenic agent approved for patients with cancer was bevacizumab, an antibody specific to vascular endothelial growth factor (VEGF), on the basis of the increased survival seen in patients with metastatic colorectal cancer with the combination of bevacizumab and standard chemotherapy in a pivotal randomized placebo-controlled phase III trial.⁹ At present, various antiangiogenesis and proangiogenesis strategies are being evaluated clinically to prevent or treat a large number of diseases, including cancer.^12–12 Both normal and pathological angiogenic processes are governed by the net balance between proangiogenic and antiangiogenic factors.^13,14 This balance is spatially and temporally regulated under physiological conditions, so that the “angiogenic switch” is “on” when needed (e.g., during embryonic development, wound healing, and formation of the corpus luteum) and “off” at other times. During neoplastic transformation and tumor progression, this regulation is deranged, and blood vessels form ectopically to support a growing tumor mass.

Cellular Mechanisms

At least four cellular mechanisms are involved in the vascularization of tumors: co-option, intussusception, sprouting (angiogenesis), and vasculogenesis (Fig. 8-2).¹¹ Tumor cells can co-opt and grow around existing vessels to form “perivascular” cuffs. However, as stated earlier, these cuffs cannot grow beyond the diffusion limit of critical nutrients and may actually cause the collapse of the vessels because of growth pressure (referred to as “solid stress”). Alternatively, an existing vessel may enlarge in response to the growth factors released by tumors, and an interstitial tissue column may grow in the enlarged lumen and partition the lumen to form an expanded vascular network. This mode of intussusceptive microvascular growth has been observed during tumor growth, wound healing, and gene therapy.^15–18 “Sprouting” angiogenesis is perhaps the most widely studied mechanism of vessel formation. During sprouting angiogenesis, the existing vessels become leaky in response to growth factors released by normal cells or cancer cells; the basement membrane and the interstitial matrix dissolve; pericytes dissociate from the vessel; endothelial cells (ECs) migrate and proliferate to form an array/sprout; a lumen is formed in the sprout (a process referred to as canalization); branches and loops are formed by confluence and anastomosis of sprouts to permit blood flow; and finally, these immature vessels are invested in basement membrane and pericytes. During physiological angiogenesis, these vessels differentiate into mature arterioles, capillaries, and venules, whereas in tumors they remain largely immature.^5,11,12,19 During embryonic development, a primitive vascular plexus is formed from endothelial precursor cells (EPCs, also known as angioblasts) by a process referred to as vasculogenesis. In adults, EPCs—mobilized from bone marrow niches into the peripheral blood circulation—also can contribute to neovascularization (a process referred to as “postnatal” vasculogenesis) in tumors and other tissues.^22–22 In addition, recent reports demonstrated that a fraction of endothelial cells in tumor vessels may be derived via transdifferentiation from cancer cells,²³ or from cancer stemlike cells (e.g., in glioblastomas).^11,24–26 In this larger context, the term “vasculogenesis” might be invoked to describe neovascularization in tumors from tumor stem cells.²⁷ Tumor vasculogenesis also may occur from EPCs of non–bone marrow origin that are recruited from adjacent tissues and/or circulation.^28,29 The current challenge is to discern the relative contribution of each of the these mechanisms of neovascularization during tumor growth and during and after treatment of tumors.^27,30

Molecular Mechanisms

Various proangiogenic and antiangiogenic molecules that orchestrate different steps in vessel formation, along with their functions, are listed in Table 8-1. VEGF currently is considered the most critical proangiogenic molecule. Originally discovered in 1983 as the vascular permeability factor and cloned in 1989, VEGF increases vascular permeability, promotes migration and proliferation of ECs, serves as an EC survival factor, can mobilize EPC populations from the bone marrow, and is known to upregulate leukocyte adhesion molecules on ECs.^{15,22,31–33} During tumor progression, or with treatment, the number of distinct angiogenic molecules produced by a tumor can increase.^36–36 Thus after VEGF signaling is blocked, a tumor might rely on other, alternative angiogenic molecules (e.g., basic fibroblast growth factor [bFGF], stromal-derived factor 1α [SDF1α], placental growth factor [PlGF], or interleukin-8 [IL-8]).³⁷ Other positive regulators of angiogenesis include the angiopoietins that are involved in stabilizing vessels and controlling vascular permeability; various proteases involved in dissolving/remodeling matrix and releasing growth factors; and recently discovered organ-specific angiogenic stimulators (e.g., endocrine gland–derived VEGF).^19,38,39 Angiogenesis inhibitors include endogenous soluble receptors of various proangiogenic ligands (e.g., sVEGFR1/sFLT1) and molecules that downregulate the expression of stimulators (e.g., interferons) or that interfere with the release of the stimulators or binding with their receptors (e.g., platelet factor 4). Thrombospondins are among the first and most well-characterized endogenous inhibitors that interfere with the growth, adhesion, migration, and survival of ECs.¹³ Other endogenous inhibitors include fragments of various plasma or matrix proteins (e.g., angiostatin, a fragment of plasminogen; endostatin, a fragment of collagen XVIII; and tumstatin, a fragment of collagen IV).^42–42 Neither the mechanisms of action of the matrix-derived inhibitors nor their physiological role are well understood.⁴³ The generation of proangiogenic and antiangiogenic molecules can be triggered by metabolic stress (e.g., low po₂, low pH, or hypoglycemia), mechanical stress (e.g., shear stress or solid stress), immune/inflammatory cells that have infiltrated the tissue, and genetic mutations (e.g., activation of oncogenes or deletion of suppressor genes that control the production of angiogenesis regulators).^{13,14,44–47} These molecules can emanate from cancer cells, endothelial cells, stromal cells, blood, and extracellular matrix (Fig. 8-3).^48–51 Because the normal host cells differ among organs, the detailed mechanisms of angiogenesis might depend on the specific host-tumor interactions operating within a given tissue.^52–59 Furthermore, because the tumor microenvironment is likely to change during tumor growth, regression, and relapse, profiles of proangiogenic and antiangiogenic molecules are likely to change with time and space.^60,61 The challenge currently is to develop a unified conceptual framework to describe the temporal and spatial profiles of this increasingly diverse array of angiogenesis regulators with the aim of developing effective therapeutic strategies.^37,62–64

Table 8-1

Angiogenesis Activators and Inhibitors*

Activators	Function	Inhibitors	Function
VEGF family members†‡	Stimulate angio/vasculogenesis, permeability, leukocyte adhesion	VEGFR-1; soluble VEGFR-1; soluble NRP-1	Sink for VEGF, VEGF-B, PlGF
VEGFR‡, NRP-1, NRP-2	Integrate angiogenic and survival signals	Ang 2†‡	Antagonist of Ang 1
EG-VEGF	Stimulate growth of endothelial cells derived from endocrine glands	TSP-1,2	Inhibit endothelial migration, growth, adhesion, and survival
Ang 1 and Tie 2†‡	Stabilize vessels	Angiostatin and related plasminogen kringles	Inhibit endothelial migration and survival
PDGF-BB and receptors	Recruit smooth muscle cells	Endostatin (collagen XVIII fragment)	Inhibit endothelial survival and migration
TGF-β1,§ endoglin, TGF-β receptors	Stimulate extracellular matrix production	Tumstatin (collagen IV fragment)	Inhibit endothelial protein synthesis
FGF, HGF, MCP-1	Stimulate angio/arteriogenesis	Vasostatin; calreticulin	Inhibit endothelial growth
Integrins αvβ3, αvβ5, α5β1	Receptors for matrix macromolecules and proteinases	Platelet factor-4	Inhibit binding of bFGF and VEGF
VE-cadherin; PECAM (CD31)	Endothelial junctional molecules	Tissue-inhibitors of MMP (TIMPs); MMP-inhibitors; PEX	Suppress pathological angiogenesis
Ephrins‡	Regulate arterial/venous specification	Meth-1; Meth-2	Inhibitors containing MMP-, TSP-, and disintegrin domains
Plasminogen activators, MMPs	Remodel matrix, release growth factor	IFN-α, -β, -γ; IP-10, IL-4, IL-12, IL-18	Inhibit endothelial migration; downregulate bFGF
PAI-1	Stabilize nascent vessels	Prothrombin kringle-2; anti-thrombin III fragment	Suppress endothelial growth
NOS; COX-2	Stimulate angiogenesis and vasodilation	16 kD-prolactin	Inhibit bFGF/VEGF
AC133	Regulate angioblast differentiation	VEGI	Modulate cell growth
Chemokines§	Pleiotropic role in angiogenesis	Fragment of SPARC	Inhibit endothelial binding and activity of VEGF
Id1/Id3	Inhibit differentiation	Osteopontin fragment Maspin Canstatin, proliferin-related protein, restin	Interfere with integrin signaling Protease inhibitor mechanisms unknown

COX-2, Cyclooxygenase-2; EG-VEGF, endocrine gland–derived vascular endothelial growth factor; bFGF, basic fibroblast growth factor; HGF, hepatocyte growth factor; IFN, interferon; IL, interleukin; IP, interferon-inducible protein; MCP-1, monocyte chemotactic protein-1; MMP, metalloproteinase; NRP-1, neuropilin-1; NOS, nitric oxide synthase; PAI-1, plasminogen activator inhibitor-1; PDGF-BB, platelet-derived growth factor-BB; PECAM, platelet endothelial cell adhesion molecule; PEX, hemopexin fragment of MMP-2; PlGF, placental growth factor; SPARC, secreted protein acidic and rich in cysteine; TGF, transforming growth factor; TIMP, tissue inhibitor of metalloproteinase; TSP, thrombospondin; VEGF, vascular endothelial growth factor; VEGFR, vascular endothelial growth factor receptor; VEGI, vascular endothelial growth inhibitor.

^*Selected list updated from reference 11; for complete function and references, see supplementary information (http://steele.mgh.harvard.edu).

^†Also present in or affecting nonendothelial cells.

^‡See reference 19.

^§Opposite effect in some contexts.

Vascular Architecture

In normal tissue, blood flows from an artery to arterioles to capillaries to venules to a vein. Although the tumor vasculature originates from these host vessels and the mechanisms of angiogenesis are similar, its organization may differ dramatically, depending on the tumor type, its location, and whether it is growing, regressing, or relapsing.15,65–68 In general, tumor vessels are dilated, saccular, tortuous, and chaotic in their patterns of interconnection.⁶⁹ For example, whereas normal vasculature is characterized by dichotomous branching, tumor vasculature has many trifurcations and branches with uneven diameters.^70,71 The fractal dimensions and minimum path lengths of tumor vasculature are different from those of normal host vasculature.^67–67 The molecular mechanisms of this abnormal vascular architecture are not understood, but it seems reasonable to hypothesize that the imbalance of VEGF and angiopoietins is a key contributor.^19,72 In mice, “normalization” of the tumor vasculature observed during therapies that reduce VEGF (e.g., hormone withdrawal from a hormone-dependent tumor), interfere with VEGF signaling (e.g., treatment with anti-VEGF or anti-VEGFR2 antibody; Fig. 8-4), or mimic an antiangiogenic cocktail (e.g., trastuzumab [Herceptin] treatment of a HER2-overexpressing tumor) is in concert with this molecular hypothesis.^{61,63,73–76} Mechanical solid stress (i.e., physical forces exerted by solid tissue components) generated by proliferating cancer and stromal cells and their excessive interstitial matrix production also can lead to the partially compressed or totally collapsed vessels often found in tumors (Fig. 8-5).^64,77,78 The decompression of blood vessels observed after induction of apoptosis in perivascular cells or stromal cells, or enzymatic degradation of matrix hyaluronan, supports this mechanical hypothesis.^78–81 Perhaps the combination of both molecular and mechanical factors renders the tumor vasculature abnormal, and thus both types of factors must be taken into account when designing novel strategies for cancer treatment.

Blood Flow and Microcirculation

Blood flow in a vascular network, whether normal or abnormal, is governed by the arteriovenous pressure difference and flow resistance. Flow resistance is a function of the vascular architecture (referred to as geometric resistance) and of the blood viscosity (rheology, referred to as viscous resistance).⁶⁹ Abnormalities in both vasculature and viscosity increase the resistance to blood flow in tumors.^71,82–84 As a result, overall perfusion rates (i.e., blood flow rate per unit volume) in tumors are lower than in many normal tissues.^85,86 Both macroscopically and microscopically, tumor blood flow is temporally and spatially chaotic. Macroscopically, four spatial regions can be recognized in a tumor (Fig. 8-6):

At the microscopic level, in normal tissues, erythrocyte velocity is dependent on vessel diameter, but there is no such dependence in most tumors.54,59,89 Furthermore, the average erythrocyte velocity can be an order of magnitude lower in some tumors compared with that of normal host tissue (Fig. 8-7).⁵⁹ In a given vessel within a tumor, blood flow fluctuates with time and can reverse its direction.^87,89,90 In addition to the elevated geometric and viscous (rheologic) resistance, other molecular and mechanical factors contribute to this spatial and temporal heterogeneity. These factors include imbalance between proangiogenic and antiangiogenic molecules, “solid stress” generated by proliferating tumor cells and matrix production, vascular remodeling by intussusception, and coupling between luminal and interstitial fluid pressure via hyperpermeability of tumor vessels.^{6,16,64,65,77–79,91–93} This heterogeneity contributes to both acute and chronic hypoxia in tumors—a major cause of resistance to radiation and other therapies. Considerable effort has gone into increasing tumor blood flow for improving radiation therapy, or decreasing tumor perfusion in the case of hyperthermia. Increased perfusion also correlates with a positive response to radio/chemotherapy and patient survival.^94,95 Increased perfusion has been difficult to achieve reproducibly, because tumor vasculature consists of both vessels co-opted from the preexisting host vasculature and vessels resulting from the angiogenic response of host vessels to cancer cells. The former are invested in normal contractile perivascular cells, whereas the latter lack these perivascular cells or these cells are abnormal.^48,69,96 Presumably as a result, efforts to increase tumor blood flow by pharmacologic or physical agents have not always been reproducible or successful.^48,69,85,96 On the other hand, the strategy of decreasing or shutting down tumor blood flow by “stealing” blood away from the “passive component” of the tumor vasculature by vasodilators, by vascular targeting, or by intravascular coagulation has shown promise in experimental systems.^{85,86,97–99} It also appears that judiciously applied antiangiogenic therapy could “normalize” the abnormal tumor microcirculation by pruning the immature vessels (see Fig. 8-4), thus rendering the remaining vasculature more responsive to vasoactive agents.¹⁰⁰

Movement of Cells Across Vessel Walls

Both cancer cells and immune cells frequently move across the walls of blood vessels—the former in the process of metastasis, and the latter during immune response or cell-based immunotherapy. Both transendothelial pathways (through ECs) and periendothelial pathways (between ECs) have been proposed as a route for intravasation and extravasation of cells. Very little is known about intravasation except that tumors might shed more than a million cells per gram per day and most of these are not clonogenic.119–122 Recently, it was demonstrated that metastatic cells could bring their own stroma from the primary tumor in the form of hereotypic cell clumps.¹²³ More is known about the molecular and cellular mechanisms of extravasation.¹²⁴ When a cell enters a blood vessel, it can continue to move with the flowing blood, collide with the vessel wall, adhere transiently or stably, and finally extravasate. These interactions are governed both by local hydrodynamic forces and adhesive forces. The former are determined by the vessel diameter and fluid velocity and the latter by the expression, strength, and kinetics of bond formation between adhesion molecules and by the surface area of contact.^125–129 Deformability of cells affects both types of forces.¹³⁰ In addition, cancer cells may grow intravascularly at a distant site (e.g., in the lungs).¹³¹

Rolling of endogenous leukocytes is generally low in tumor vessels, whereas stable adhesion (≥30 sec) is comparable between normal vessels and tumor vessels.¹³² On the other hand, both rolling and stable adhesion are nearly zero in angiogenic vessels induced in collagen gels by bFGF or VEGF, two of the most potent angiogenic factors.¹³³ Whether this observation is due to a low flux of leukocytes into angiogenic vessels and/or to downregulation of adhesion molecules in these immature vessels currently is not known. The age of the animal also plays an important role in leukocyte-endothelial interactions.¹³⁴ Further insight into the types of cells that adhere to tumor vessels comes from studies on the localization of IL-2-activated natural killer (A-NK) cells in normal and tumor tissues in mice using positron emission tomography.^135,136 After systemic injection, these cells localized primarily in the lungs immediately after injection and could not be detected in the tumor.¹³⁵ Increased rigidity caused by IL-2 activation might contribute to the mechanical entrapment of these cells in the lung microcirculation.^137,138 Constitutive expression of certain adhesion molecules in the lung vasculature also might facilitate their retention in the lungs.¹²⁴ One approach to reducing lung entrapment is to reduce the rigidity of these cells.^112,117 Alternatively, the lung can be circumvented by injecting A-NK cells directly into the blood supply of tumors. In this case, A-NK cells, both xenogenic and syngeneic, adhered to some blood vessels in three different tumor models via CD18 and very large antigen-4 (VLA-4) on the A-NK cells and intercellular adhesion molecule-1 (ICAM-1), vascular cell adhesion molecule-1 (VCAM-1), and E-selectin on the activated endothelium of angiogenic vessels.^{33,136,139–142} These molecules can be upregulated by tumor necrosis factor–α (TNF-α) and a protein of 90 kd molecular weight (p90) that is secreted by some neoplastic cells and downregulated by transforming growth factor–β (TGF-β)—also, presumably, secreted by cancer cells.^{55,125,143–147} Surprisingly, the proangiogenic VEGF also can upregulate these molecules, whereas another proangiogenic molecule, bFGF, can downregulate these molecules.^{33,51,61,76,148} Inflammatory cells such as monocyte/macrophages or neutrophils also are recruited by VEGF and may play important roles in promoting matrix remodeling and angiogenesis.¹⁴⁹ The challenge currently is to decrease nonspecific entrapment of immune cells in normal vessels and to increase their delivery to tumor vessels to improve various cell-based therapies, including gene therapy.

Interstitial Transport

Once a molecule has extravasated, its movement through the interstitial space occurs by diffusion and convection.101,154 Diffusion is proportional to the concentration gradient in the interstitium, and convection is proportional to the interstitial fluid velocity, which, in turn, is proportional to the fluid pressure gradient in the interstitium. Just as the interstitial diffusion coefficient D (cm²/s) relates the diffusive flux to the concentration gradient, the interstitial hydraulic conductivity K (cm²/mmHg/sec) relates the interstitial velocity to the pressure gradient.¹³⁴ Values of these transport coefficients are governed by the structure and composition of the interstitial compartment and by the physicochemical properties of the solute molecule.^155–165 The value of K for a human colon carcinoma xenograft (LS174T), measured using two different methods, was found to be higher than that of a hepatoma, which, in turn, was higher than that of the normal liver.^167–167 Using fluorescence recovery after photobleaching, D of various molecules in tumors was found to be about one third that in water and higher than the values in the host tissue (Fig. 8-9, A).^157,168 Collagen content and structure have a significant effect on D in tumors.^{159,163,166,169–173} This finding is surprising because hyaluronan and proteoglycans, not collagen, account for most of the resistance to transport in normal tissues. Because collagen is produced by host cells (e.g., fibroblasts), the penetrability into a tumor depends on the host-tumor interaction (see Fig. 8-9, A). Thus agents that interfere with collagen synthesis and/or organization (e.g., relaxin, losartan, and bacterial collagenase) might increase interstitial transport in tumors^156,169,170 (Fig. 8-9, B). The time constant for a molecule with diffusion coefficient D to diffuse across a distance L is approximately L²/4D. For diffusion of immunoglobulin G (IgG) in tumors, this time constant is on the order of 1 hour for a 100-µm distance, days for a 1-mm distance, and months for a 1-cm distance. Thus for a 1-mm distance in tumor, diffusional transport would take days, and for a 1-cm distance in tumor, it would take months. Furthermore, the path through the interstitium is tortuous over large length scales, greatly increasing the distance a molecule must travel.¹⁵⁸ If the central vessels have collapsed completely as a result of cellular proliferation and interstitial matrix rearrangement, the reduced delivery of macromolecules by blood flow would make diffusion the primary mechanism of delivery to this necrotic center.^6,79 Binding of a low- or high-molecular-weight drug to plasma proteins and various tissue components could further retard their transport in tumors.^{168,174–180} The role of binding is clearly illustrated in Figure 8-9, C, which compares the rate of fluorescence recovery of a photobleached spot in tumor tissue injected with a nonspecific versus a specific IgG. In addition to the heterogeneity of D in tumors, the most unexpected result of these photobleaching studies was the large extent (30% to 40%) of nonspecific binding.¹⁶⁸ These results collectively suggest that the interstitial compartment of a tumor can be a formidable barrier to the uniform delivery of therapeutic macromolecules (e.g., antibodies and genes) in tumors, and strategies are needed to modify this barrier.

Lymphangiogenesis and Lymphatic Transport

In most normal tissues, extravasated plasma and macromolecules are taken up by the lymphatics and returned to the blood circulation. It is widely accepted that lymphatic vessels are present in the tumor margin and the peritumoral tissue (Fig. 8-10, A). Indeed, invasion of peritumoral lymphatics is considered to be a poor prognostic factor for several tumors (e.g., breast, colorectal, and endometrial cancers), and lymphatic metastasis is a major cause of morbidity and mortality for others (e.g., melanoma, head and neck cancer, lung cancer, and cervical cancer). The hotly debated issue for nearly a century has been whether anatomically defined lymphatic vessels are present within solid tumors and, if so, whether they function (see Fig. 8-10, A).^181,182 Currently available immunohistochemical markers stain for structures in some tumors that resemble lymphatic vessels. Because many of these markers lack specificity, however, it is not clear whether they stain functional lymphatic vessels, ECs from remnant lymphatic vessels, or some other structures (e.g., preferential fluid channels).^{152,166,183,184} Mechanical “solid stress” induced by proliferating tumor cells and matrix production compresses and impairs lymphatic vessels that are co-opted or formed within a tumor.^6,78,80 The impaired lymphatic vessels, in turn, contribute to the interstitial hypertension characteristic of animal and human tumors (to be discussed shortly). Embryonic lymphatic vessels originate primarily from blood vessels according to the following process (Fig. 8-10, B).^187–187

Members of the angiopoietin family (Ang-2) and its receptor (Tie-2), as well as podoplanin, presumably are involved in the maturation and patterning of these nascent lymphatic vessels.¹⁸⁸ The hematopoietic signaling pathway, Syk/SLP-76, contributes to the separation of lymphatics from blood vessels. The molecules involved in angiogenesis also are involved in lymphangiogenesis. For example, VEGF-C and VEGF-D can induce both angiogenesis and lymphangiogenesis and are associated with lymphogenic metastasis in a variety of tumors.¹⁵¹ Their receptor VEGFR3 is present in both lymphatic and selected vascular endothelium. As is the case with vascular angiogenesis, other positive and negative regulators (e.g., angiopoietins) and other receptors (e.g., chemokine receptors and neuropilins) could be involved in lymphangiogenesis, and as previously discussed, mechanisms analogous to co-option, intussusception, sprouting, and vasculogenesis might operate in lymphatic growth.^11,188 Similar to the recently discovered organ-specific angiogenic molecule (EG-VEGF) and endothelial precursor cells, there could be organ-specific lymphangiogenic molecules and lymphatic EPCs that contribute to tumor-associated lymphangiogenesis.^22,38,186 Moreover, the proteolytic processing of lymphangiogenic molecules and the phenotype and function of the resulting lymphatics might depend not only on the tumor type but also on the host organ in which the tumor is growing.^76,184–186

The mechanical and/or molecular signals that could trigger the lymphangiogenic switch are unknown. Because lymphatic vessels help maintain the balance of fluid in tissues, hydrostatic pressure is a probable trigger. Whether the hyperplasia and the increased density of lymphatic vessels seen in the tumor margins are a response to elevated hydrostatic pressure in tumors and whether the newly formed lymphatics are able to remain open and carry cancer cells are open questions. Techniques such as microlymphangiography, reagents that block signaling of VEGF-C and VEGF-D, and lymphangiogenic factors yet to be discovered will allow us to answer these important questions.4,184,186,189–197

Interstitial Hypertension

Unlike normal tissues, in which the interstitial fluid pressure (IFP) is around 0 mm Hg, both animal and human tumors exhibit interstitial hypertension (Table 8-2).^{151,154,167,198–209} The tumor IFP begins to increase as soon as the host vessels become leaky in response to permeability factors such as VEGF, and thus IFP can be lowered by antibodies against VEGF or VEGFR2.^212–212 The IFP increases with tumor size in some tumors and remains independent of tumor size in others.^{198–200,202,203} Three mechanisms contribute to interstitial hypertension in tumors. In normal tissues, the lymphatics maintain the fluid homeostasis; thus the lack of functional lymphatics in tumors is a key contributor. Indeed, DiResta and colleagues have been able to lower the IFP by placing “artificial lymphatics” in tumors.²¹³ The second contributor is the high permeability of tumor vessels. As a result, the hydrostatic and oncotic (colloid osmotic) pressures become almost equal between the intravascular and extravascular spaces.^201,214 At least two pieces of evidence support this hypothesis. First, reducing permeability by blocking VEGF signaling lowers IFP.^{211,212,215–218} Second, IFP increases and decreases with the microvascular pressure within seconds.^221–221 The two mechanisms described thus far can only explain interstitial hypertension up to 20 to 30 mm Hg, the microvascular pressure of most exchange vessels (the hydrostatic pressure within the lumina of capillaries) in our bodies, but IFPs as high as 94 mm Hg have been measured in human tumors.²⁰⁰ Because microvascular pressure (MVP) is the driving force for IFP in tumors, these tumors must have a high MVP. Indeed, this is the case.²⁰¹ Two possible explanations exist for elevated MVP in tumors: the tumor vessels have reduced arterial resistance so that the MVP becomes closer to arterial pressure, and/or the tumor vessels have increased venous resistance as a result of compression and tortuosity so that the entire intratumor vascular network is under hypertension. Indirect evidence for the latter explanation comes from the decrease in IFP after decompression of tumor vessels by taxol-induced apoptosis of perivascular cells.⁷⁹

The elevated pressure can compromise the tumor microcirculation and delivery of therapeutics in three ways:

1. Reduced transmural pressure gradients due to equilibrium between MVP and IFP reduce convection across tumor vessel walls and thus compromise the transport of macromolecules.201,220

2. Because IFP is nearly uniform throughout a tumor and drops precipitously in the tumor margin, the interstitial fluid oozes out of the tumor into the surrounding normal tissue, carrying macromolecules with it (Fig. 8-11).^198,222

3. Finally, transmural coupling between IFP and MVP due to high permeability of tumor vessels can lead to blood flow stasis in tumors without physically occluding the vessels.93–93

The enhanced permeability also can facilitate lymphatic metastasis (Fig. 8-10, C). Thus decreasing vascular permeability might restore the transmural pressure gradients and potentially resume/reestablish blood flow in the nonperfused regions of tumors. Some direct and indirect antiangiogenic therapies might “normalize” the tumor vasculature through this mechanism (see Fig. 8-4).¹⁰⁰ This normalization also can reduce the number of cells shed into peritumor lymphatics and alleviate peritumor edema (see Fig. 8-10, C).^73,223

Low pH

Another consequence of the abnormal microcirculation of the tumor is low extracellular pH. There are at least two sources of H⁺ ions in tumors—lactic acid and carbonic acid.²³⁰ The former results from glycolysis and the latter from conversion of CO₂ and H₂O via carbonic anhydrase. The intracellular pH of cancer cells remains neutral or alkaline (pH 7.4), however, despite the acidic extracellular pH. Because carbonic anhydrase-9 (CAIX), various glucose transporters (GLUT1, GLUT3), and enzymes in the glycolytic pathway are upregulated by hypoxia, one would expect low extracellular pH and hypoxia to track each other and to colocalize with regions of low blood flow.²³¹ Surprisingly, there is a lack of spatial correlation among these parameters (Fig. 8-12, B), a discovery made possible by recent developments in optical techniques that permit the simultaneous high-resolution mapping of multiple physiologic parameters.²²⁶ A potential explanation for this lack of concordance is that some perfused tumor vessels carry hypoxic blood.²²⁶ Thus although they might not be able to deliver adequate oxygen to the surrounding cells, they may be able to carry away the waste products (e.g., lactic acid).

Molecular, Cellular, and Therapeutic Consequences

The presence of oxygen during irradiation makes the damage to DNA produced by radiation-induced free radicals permanent, whereas such damage can be repaired under hypoxic conditions.²³² Therefore hypoxia in solid tumors significantly reduces their radiation sensitivity. Tumor hypoxia also is associated with resistance to some chemotherapeutics.²³² Similarly, low extracellular pH can affect the cellular uptake and cytotoxicity of some therapeutics adversely or favorably.^235–235 As a result, for nearly half a century, considerable preclinical and clinical efforts have been focused on alleviating hypoxia by improving tumor perfusion with various therapies, including the following:

Unfortunately, the clinical outcome has not met expectations for multiple reasons. These reasons include the inability to increase po₂ in all regions of tumors to optimal levels and/or to deliver radiation sensitizers or chemotherapeutic drugs to all regions of a tumor at therapeutically effective levels. As a result, three broad strategies are emerging:

The first strategy has led to the development of agents such as tirapazamine and to the rejuvenation of interest in bacteriolytic therapy; both approaches are still being tested in clinical trials.88,232 The second strategy has revealed several molecular determinants of the physiological and pathophysiological responses to hypoxia.²³¹ The balance between hypoxia-induced apoptosis/necrosis on one hand, and the increased resistance to cell death mediated by various hypoxia-induced pathways on the other, determines whether a tumor can survive and even grow under hypoxic conditions. Ultimately, hypoxia might select for tumor cells that are more malignant, more invasive and genetically unstable, and less susceptible to apoptosis, thus rendering them resistant to various therapies. Therefore several molecules in the hypoxia-induced pathways are now being targeted in the development of diagnostic and therapeutic agents. Finally, normalization of tumor vessels by antiangiogenic agents may reduce hypoxia within tumor environment and synergize with radiation therapy.²³⁷

Hypoxia-induced pathways include genes involved in the following processes:

Of the various molecules involved in sensing and responding to hypoxia, HIF1α has received the most attention. This transcription factor is upregulated in several human tumors.²³¹ Regulated by a proline hydroxylase, HIF1α can activate the genes for angiogenesis, vasodilation, glycolysis, and erythrocyte production by binding to the hypoxia-response element.^238,239 Surprisingly, teratomas arising from HIF1α^–/– embryonic stem cells grow more rapidly despite lower levels of VEGF and angiogenesis.^51,240 This counterintuitive finding could be a result of the ability of HIF1α^–/– cells to survive under hypoxic conditions instead of undergoing apoptosis.⁴⁸ Interestingly, other HIF1α^–/– cancer cells lead to slowly growing tumors. As a result, molecular therapies that target HIF1α or hypoxia-response element are under intensive investigation for cancer detection and treatment.²⁴¹

Mechanisms of Action

Recent clinical data offer hope, with seven U.S. Food and Drug Administration–approved VEGF-blocking antiangiogenic agents (bevacizumab, sorafenib, sunitinib, pazopanib, axitinib, vandetanib, regorafenib, and aflibercept) already in clinical use.11,245 Studies in patients with rectal cancer and glioblastoma have largely confirmed the hypotheses that anti-VEGF therapy has antivascular effects and can induce vascular normalization in patients with cancer.^{216,218,245–249} In brief, bevacizumab alone reduced the tumor tissue vascular density by approximately half at day 12 after the first infusion and reduced significantly the tumor blood flow evaluated on computed tomography scans and the number of viable circulating endothelial cells and progenitor cells. Clinically, both the low-dose-bevacizumab infusion and the high-dose-bevacizumab infusion showed a less hyperemic/hemorrhagic appearance but no significant tumor regression after bevacizumab treatment upon flexible sigmoidoscopy.^216,218 Although the significant pruning of tumor vasculature led to a significant increase in cancer cell apoptosis, it also led to a more mature (pericyte-covered) tumor vasculature and a stable or increased cancer cell proliferation (Fig. 8-13, A-D).^218–218 Whether the increase in tumor cell apoptosis was due to a direct or indirect effect of bevacizumab is currently unclear. Similarly, the role of VEGF blockade on immune or stromal cells in patients is not yet understood. After bevacizumab therapy alone, the tumor IFP was consistently decreased, particularly in the patients with high baseline values.^218–218 This finding suggested that in human tumors, similar to mouse models, the tumor microenvironment was normalized by the reduction of the excessive vascularization and potentially sensitized the tumor to the subsequent cytotoxic therapy within the “normalization window.”^212,237 Imaging studies yielded more supportive data for the normalization hypothesis: despite the significant reduction in vessel density and blood flow, the fluorodeoxyglucose uptake measured on positron emission tomography scans (a measure of tumor metabolic activity) and the P • S product (proportional to the penetration of tracer in tumor) evaluated on computed tomography scans did not significantly change at day 12.^216,218 Antiangiogenic therapy with cediranib in recurrent glioblastoma demonstrated the existence of a window of vascular normalization in patients with cancer (Fig. 8-13, E and F).^247,248 Moreover, two recent studies of cediranib in glioblastoma showed that some patients respond with increased perfusion and that these patients survive longer than do nonresponders.⁹⁵

Biomarkers

Similar to many targeted therapies, the survival benefit after antiangiogenic therapies remains modest. Unfortunately, unlike many anticancer targeted therapies, there are no validated biomarkers of patient selection, response, or resistance for antiangiogenic agents.³⁷ If we could we find biomarkers to identify patients more likely to benefit from these drugs, the survival benefit in patients taking anti-VEGF drugs would be improved. Unlike preclinical models, the phase III bevacizumab experience in patients with metastatic colorectal cancer did not identify P53, KRAS, or BRAF status, VEGF or thrombospondin-2 (TSP2) expression, or microvascular density at baseline as predictive markers of response.^250,251 Similarly, circulating VEGF levels—although of potential biomarker value—have unclear predictive biomarker value.^37,252 In collaborative studies at Massachusetts General Hospital, these questions have been examined in more than 20 multidisciplinary trials by measuring tissue, circulating and imaging biomarkers at various time points during and after treatment and correlating these with treatment outcomes. Several promising biomarkers have emerged. One is the level of circulating sVEGFR1/sFLT1 (an endogenous blocker of VEGF and PlGF) as a biomarker of intrinsic resistance to anti-VEGF therapy. It was reported that patients with locally advanced rectal carcinoma who had high plasma sVEGFR1 levels before treatment were less likely to benefit—or experience toxicities—from bevacizumab therapy combined with chemoradiation.^218,253 The same association was found for newly diagnosed patients with glioblastoma multiforme (GBM), triple-negative breast cancer, hepatocellular carcinoma (HCC), and metastatic colorectal carcinoma who were treated with anti-VEGF agents.^254–257 Of interest, a retrospective analysis in two randomized phase III trials recently has shown that a genetic variation in VEGFR1 gene correlates with increased VEGFR1 expression and poor outcome of bevacizumab treatment in patients with metastatic renal cell carcinoma and pancreatic ductal adenocarcinoma.²³⁸ A second potential biomarker is the SDF1α/CXCR4 axis as a potential evasive pathway. It was found that circulating levels of the chemokine SDF1α increase in patients at the time of evasion from anti-VEGF therapies. This was the case in patients with rectal carcinoma who were treated with bevacizumab, in patients with GBM who were treated with cediranib, in patients with HCC who were treated with sunitinib, and in patients with sarcoma after sorafenib.^{215,247,256,258,259} Interestingly, the sources of SDF1α are different in each of these diseases, and so is the role of the SDF1α/CXCR4 pathway.²⁶⁰ For example, in persons with GBM this pathway appears to facilitate invasion of cancer cells and co-option of host vessels by invading cancer cells. These biomarker candidates need to be tested prospectively. Of note, based on these findings, Dr. Patrick Wen (Dana Farber Cancer Institute, Boston, MA) has started a clinical trial with AMD3100 (an anti-CXCR4 drug) plus bevacizumab in patients with recurrent GBM (ClinicalTrials.gov Identifier: NCT01339039).

Toxicity

Experimental studies have shown that, in 11 of the 17 healthy organs studied, VEGF blockade can significantly decrease the number of normal capillaries.²⁶¹ In patients with cancer, most anti-VEGF agents often induce proteinuria, hypertension, thyroid-stimulating hormone elevation, and gastrointestinal toxicity, but agent-specific toxicities also have been reported.^10,262 In addition, the long-term effects of antiangiogenic therapies in patients with less advanced lesions remain to be established.

Perspective

The major directions for the immediate future are further understanding of the mechanisms of action and identification of the first biomarkers for anti-VEGF therapy.³⁷ Achieving these goals would allow optimization of treatment protocols and reduction of the adverse effects.

First, identifying the vascular “normalization window” in patients would allow synergistic combinations with chemotherapeutics or radiotherapy. Second, understanding the mechanisms of vessel pruning and cancer cell apoptosis induced by anti-VEGF therapy, and tumor escape from it, may allow further sensitization of tumor cells to cytotoxic therapies. Third, characterization of the effect of anti-VEGF therapy on bone marrow–derived cells’ contribution to tumor growth and relapse would allow judicious and more effective approaches to therapy involving these cells. Finally, clarification of other mechanisms involving the immune system or the stromal and interstitial matrix compartments would contribute to establishing more efficacious combinatorial strategies.

In this respect, new biomarkers and improved imaging techniques will play a major role in monitoring the effects and stratifying the patients with the ultimate goal of individualized therapy.