Transvascular Drug Delivery in Solid Tumors
`
`Fan Yuan
`
`The microvessel wall
`the delivery of
`is a barrier for
`various therapeutic agents to tumor cells Tumor mi
`crovessels are in general more permeable to macromol
`ecules than normal vessels The hyperpermeability
`presumably due to the existence of large pore struc
`tures in the vessel wall induced by various cytokines
`The cutoff pore size is tumor dependent as determined
`The vascular
`by transport studies of nanoparticles
`permeability is heterogeneous in tumors and depen
`dent on physicochemical properties of molecules as
`well as the ultrastructure of the vessel wall The ultra
`
`is
`
`structure is dynamic and can be modulated by the
`tumor microenvironment The microenvironment
`itself
`can be altered by the transvascular
`transport because
`the transport may facilitate angiogenesis reduce blood
`hypertension in tumors
`flow and induce interstitial
`transport need to address mecha
`Future studies of
`nisms of the barrier formation and emphasize develop
`ment of novel strategies for circumventing or exploiting
`the vascular barrier
`
`Copyright© 1998 by VV Saunders Company
`
`agents to tumor cells in
`of therapeutic
`vivo encounters three major problems 1 me
`Delivery
`tabolism and clearance of drugs in the body 2
`physiological barriers for transport of
`therapeutic
`agents from sites of administration to tumor cells
`and 3 drug resistance of tumor cells The issue of
`for two
`drug delivery to solid tumors is unique
`reasons First anticancer
`drugs are toxic to both
`tumor and normal cells Hence the dose of drugs is
`limited by normal
`tissue tolerance In some cases
`in humans than in experimental
`tolerance is lower
`efficacy ob
`animals23 Therefore the therapeutic
`served in rodents bearing human tumor xenografts
`may not be achievable in patients owing to inad
`equate drug delivery25 Second drug delivery
`in
`tumors is nonuniform there are regions in tumors
`where the drug exposure is insufficient The hetero
`to the incom
`geneous distribution may contribute
`plete eradication of tumors by therapeutic
`agents In
`general heterogeneous distribution of small drugs
`with short plasma halflife is attributed to the chaotic
`vasculature and microcirculation in tumors whereas
`the heterogeneous
`delivery of macromolecules
`or
`nanoparticles eg liposomes viral vectors is likely
`due to heterogeneous angiogenesis as well as transvas
`cular and interstitial
`transport461012
`Specific problems in tumor
`include 1 low convective transport because of the
`lymphatics 2 outward gradient of the interstitial
`fluid pressure which may cause convective
`
`interstitial
`
`transport
`
`interstitial
`
`hypertension
`
`and the lack of functional
`
`transport
`
`the
`
`This work was supported
`in part by a Career Development Awardfrom
`Specialized Program ofResearch Excellence SPORE in Breast Cancer
`at Duke University P50CA 6843802
`to Fan Yuan PhD Department of Biomedical
`Engineering Box 90281 Duke Universibt Durham NC 27708
`Copyright © 1998 by WB Saunders Company
`
`Address
`
`reprint
`
`requests
`
`10534296198108030003880010
`
`of extravasated drugs from the interior to the periph
`ery of tumors 3 large diffusion distance in some
`regions and 4 binding of drugs to tumor and
`
`the interstitial
`
`stroma cells as well as to the extracellular matrix5
`Review of
`
`in tumors is
`transport
`beyond the scope of this article for review see Jain5
`Therefore the following discussion is focused on the
`transvascular
`transport of therapeutic agents in solid
`
`tumors
`
`Transvascular Transport in Tumors
`
`Transvascular
`
`vessel
`
`transport in tumors is heterogeneous
`This heterogeneity
`is exemplified by examining lipo
`in tumors When injected into the
`somal
`transport
`systemic circulation fluorescently labeled liposomes
`accumulate in certain regions in solid tumors but are
`in others Fig 112 Even along the same
`absent
`the distribution of liposomes can be nonuni
`form Fig 1 Mechanisms
`of
`the heterogeneous
`transport are multifactorial and not well understood
`Transvascular
`is characterized
`by the
`transport
`to water and the microvascu
`hydraulic conductivity
`lar permeability to other molecules Both of them are
`quantities for characterizing mo
`phenomenological
`lecular transport across membranes The hydraulic
`conductivity of tumor vessels his not yet been quanti
`fied The
`capillary filtration coefficient however
`which is the product LpS of the hydraulic conductiv
`ity Lp and the surface area of vessels S has been
`
`reported in the literature13 The filtration coefficient
`in an isolated rat mammary adenocarcinoma
`R3230AC perfused ex vivo is much higher than
`that in normal tissues13 It
`is likely that the hydraulic
`in tumors is also higher than normal
`conductivity
`tissues as indicated indirectly by the vasogenic cere
`bral edema in brain tumor patients14 and the ele
`fluid pressure in most solid tumors5
`The microvascular
`permeability in tumors has
`
`vated interstitial
`
`164
`
`Seminars in Radiation Oncology Vol 8 No 3 July 1998 pp 164175
`
`Abraxis EX2047
`Actavis LLC v Abraxis Bioscience LLC
`1PR201701101 1PR201701103 1PR201701104
`
`

`

`Tranrvascular Drug Delivery
`
`in Solid Tumors
`
`165
`
`11111111111
`
`SWIM
`
`cells
`
`Figure 1 Heterogeneous
`tissues Human colon adenocareinoma
`in tumor and normal
`distribution of liposomes
`IS1741 were transplanted in dorsal skinfold chambers in severe combined immunodeficient mice Fifteen to 32 days after
`labeled liposomes were injected intravenously The photos were taken 2 days after
`tumor cell transplantation fluorescently
`injections A Local heterogeneity Lijxmornes accumulated
`regions in solid tumors Bar 1001trit B
`only in perivaseular
`Regional heterogeneity There was a significant extravasation of liposomes on the left and nearly no extravasationon the
`and well perfused Bar 400 pal C Most
`right Both sides of the tumor were vaseularized
`liposomes
`extravasated
`near the roots of capillary sprouts whereas the sprouts per se showed minimal leakiness Bar 200 Am D
`accumulated
`Liposomes accumulated only in the wall of small postcapillary venules to 25 Arn in diameter in normal subcutaneous
`by arrows Neither parallel capillaries e nor arterioles a and large collecting
`venulea v >25 pm
`were labeled by liposomes Bar 100 pm Reprinted with permison12
`MW = 376 and I assamine green MW = 57718 In
`the clinic computed tomography and magnetic reso
`nance imaging studies demonstrate that some hu
`man glioblastomas
`and cerebral
`lymphomas do not
`show contrast enhancement after infusion ofcontrast
`agents and the percentage of contrast enhance
`ment may depend on the type stage and location of
`tumors as well as the age of patients Ultrastruc
`tural studies of human brain tumors also reveal
`vessels arc ob
`results Fenestrated
`heterogeneous
`served in some ghat tumors 29 but not in others3° For
`nonglial tumors and brain metastasis the results are
`more consistent vessels in these tumors are fenes
`trated3 The existence of fenestrated
`vessels
`in
`solid tumors however has been challenged by ani
`mal studies243334 as discussed later
`In addition to tumor vessel wall the BTB includes
`
`tissues as indicated
`
`been studied extensively The data from the litera
`ture as summarized in Table 1 demonstrate that
`tumor mierovascular permeability is in general ele
`vated Fig 051012k523 This is presumably due to the
`exposure of tumor vascular endothelial cells to cyto
`factor
`kines such as vascular endothelial growth
`vascular permeability factor VEGFVPF54°24 Pri
`mary brain tumors however may represent
`a case
`different from other tumors
`Vessels in some but not all primary brain tumors
`are nearly impermeable to therapeutic
`drugs or
`agents The tight blood tumor barrier
`diagnostic
`151B has been observed both clinically and experi
`mentally For example vessels in a human glioblas
`tonia xertograft HM21 transplanted in mice have
`been shown to be similar to the blood brain barrier
`BBB that
`is impermeable to sodium fluorescein
`
`

`

`166
`
`Fan Yuan
`
`Table 1 The Microvascular Permeability of Tumor Vessels
`
`Tumor Tissue
`
`Host Tissue
`
`Tracer
`
`Size of
`Tracer
`
`Permeability
`107 cms
`
`Reference
`
`Human colon aca LS174T
`
`SCIDt sc tissue
`
`SCID liver
`SCID pia mater
`
`SCID pia mater
`SCID pia mater
`SCID pia mater
`
`C3H sc tissue
`C3H pia mater
`SCID pia mater
`
`Human glioblastoma
`HGL2 1
`Human glioblastoma U87
`Human melanoma PMEL
`Mouse mammary aca
`MCaIV
`
`Rat mammary aca
`R3230AC
`
`Fe fragment
`Fab fragment
`Ovalbumin
`
`Albumin
`Concanavalin A
`Fab2 fragment
`IgG
`
`Stabilized liposome
`Albumin
`
`Albumin
`
`Albumin
`
`Albumin
`
`Albumin
`
`Albumin
`
`Albumin
`Albumin
`
`25000
`
`25000
`
`45000
`
`66000
`
`104000
`
`110000
`
`43 ± 16
`46 ± 10
`59 ± 12
`14 ± 05
`19 ± 07
`14 ± 03
`26 ± 11
`160000
`90 nm 020 ± 016
`48 ± 35
`66000
`45 ± 09
`
`66000
`
`66000
`
`66000
`
`66000
`
`66000
`
`66000
`66000
`
`011 ± 005
`38 ± 12
`10 ± 01
`
`21 ± 07
`29 ± 15
`31 ± 05
`
`Rat pia mater
`Rabbit mammary aca V2
`Rabbit granulation tissue
`The size of tracers is indicated by either the molecular weight without unit or the diameter nm
`and C3H are two strains of mice
`tSCID severe combined
`immunodeficient
`Abbreviations aca adenocarcinoma Sc subcutaneous
`
`Rat granulation tissue
`
`Sulforhodamine B
`Albumin
`
`558
`
`Stabilized liposome
`Conventional liposome
`Albumin
`
`Dextran
`
`340 ± 70
`78 ± 12
`66000
`82 nm 34 ± 08
`91 nm
`18 ± 04
`17 ± 06
`66000
`57 ± 39
`
`150000
`
`19
`
`19
`
`19
`
`19
`
`19
`
`19
`
`19
`
`12
`
`23
`
`20
`
`18
`
`18
`
`20
`
`22
`
`18
`20
`
`16
`
`16
`
`10
`
`10
`
`18
`
`15
`
`the wall of normal vessels
`in surrounding tissues
`tumor cells often invade normal
`This is because
`tissues and the permeability of normal vessels is low
`especially in the brain3536 Therefore these tumor
`cells can be protected from systemic drug treatment
`tumor
`and are responsible partially for the local
`recurrence
`The vascular permeability in both tumor and
`normal
`tissues depends on physicochemical proper
`ties of drugs and the ultrastructure of
`the vessel
`wal153738 which in turn can be modulated by tissue
`microenvironment5 Mechanisms
`of heterogeneous
`transport and the formation of
`transvascular
`the
`tight BIB remain the subject of ongoing investiga
`tions Several key issues regarding the transvascular
`transport are discussed
`
`for
`
`Pathways of Transvascular Transport
`the transvascular
`Several potential pathways
`transport of molecules and nanoparticles
`identified for review see Renkin39 Mechanistic
`transport of macromol
`explanations for transvascular
`ecules and nanoparticles
`in tumors however are still
`controversial The main point in the argument is not
`
`have been
`
`which pathways are available but which ones are the
`dominant channels for the transvascular
`transport
`There are two competing hypotheses One of them
`proposed by Dvorak et a1243334
`the
`suggests that
`is the interconnected
`vesiculovacu
`major pathway
`olar organelles VVOs The number of VVOs can be
`up regulated by VEGFVPF and other vasoactive
`agents the size of VVOs ranges from 50 to 415 nm in
`diameter° VVOs are separated by diaphragms the
`opening and closing of the diaphragms regulate the
`rate of transport° The second hypothesis suggests
`that
`the major pathway
`is the open endothelial
`These structures
`can be
`junctionfenestra30324143
`induced by VEGFVPF and other cytokines and are
`up to 700 nm in width4142 Despite the differences
`discussed both hypotheses agree that the transvascu
`and nanoparticles
`lar pathways for macromolecules
`are channel like structures The discrepancy between
`the VVO and the open junctionfenestra hypotheses
`is unlikely caused by differences in tumors used in
`the studies Other issues regarding the experimental
`design and data interpretation have to be addressed
`in future studies
`The identification of transport pathways has di
`
`

`

`Transvascular Drug Deliveg in Solid Tumors
`
`167
`
`rect
`
`implication in drug delivery to solid tumors
`because
`the development
`of novel strategies
`for
`modifying tumor and normal vascular permeabilities
`relies on mechanisms of transvascular
`transport
`the vascular permeability can be modulated differen
`tially between tumor and normal tissues the specific
`ity of drug delivery to tumors will be improved
`
`If
`
`significantly
`
`Cutoff Pore Size
`
`also exist
`
`Both structural and functional analyses indicate that
`large pores exist in tumor vessels that are permeable
`and
`and nanoparticles5520344245
`to macromolecules
`the cutoff size of pores is dependent on tumor and
`organ environment1°181943
`For example vessels in
`a human glioblastoma HGL21 transplanted in
`mouse cranial windows
`been shown to be
`have
`impermeable to small molecules molecular weight
`<600 indicating that
`the cutoff pore size in this
`model is smaller than 1 nm18 The cutoff pore size
`however
`in a mouse mammary adenocarcinoma
`MCaIV transplanted in dorsal skinfold chambers is
`between 1200 and 2000 nm as determined by extrava
`sation studies of liposomes with different sizes43 The
`cutoff pore size in other tumors has been reported to
`be between 100 nm and 800 nrn10121943 Large pores
`of normal
`in discontinuous
`endothelia
`sinusoids in the reticuloendothelial system RES39
`The cutoff size of pores in liver sinusoids is approxi
`mately 100 nm46
`The size and the number of pores in the vascular
`endothelium can be altered via local application of
`various
`endothelial growth factors or vasoactive
`agents414247 Alternatively neutralization or elimina
`tion of endothelial growth factors can significantly
`reduce tumor vascular permeability and pore cutoff
`is likely that
`size2043 Therefore it
`large pores in
`tumor vessels are induced and maintained by growth
`factors and other cytokines released by tumor and
`stroma cells
`Large pores in tumor vessels
`have provided
`a
`therapeutic window for specific drug delivery to solid
`tumors For example the heart microvessels are
`permeable only to molecules smaller than or similar
`peroxidase approximately 5 nm in
`to horseradish
`10 to 20 nm apart
`diameter although gaps spaced
`can also be found occasionally
`in these vessels48
`Therefore one would expect
`the therapeutic
`that
`than 20 nm and smaller than the
`agents larger
`cutoff size of pores in tumor vessels will accumulate
`preferentially in tumors and a few normal organs eg
`the RES This is indeed the case in the delivery of
`
`doxorubicin to solid tumors in patients1932 When
`doxorubicin is encapsulated
`in sterically stabilized
`liposomes of approximately 100 nm in diameter the
`severe cardiotoxicity caused by the treatment with
`free doxorubicin can be completely eliminated493132
`In addition these stabilized liposomes enhance drug
`delivery to solid tumors in comparison with conven
`liposomes The enhancement
`is attributed to
`prolonged plasma halflife reduced uptake of these
`particles in the RES and enhanced vascular perme
`ability103033 The increase in drug delivery however
`may not necessarily enhance the efficacy of drugs
`Therefore the clinical outcome
`the liposome
`of
`mediated cancer chemotherapy
`remains to be deter
`mined
`
`tional
`
`Effect of Physicochemical Properties of
`Drugs on the Microvascular Permeability
`
`Physicochemical properties of drugs such as charge
`size configuration and polarity may affect the trans
`the vessel wal1337 In general
`the
`port across
`permeability of both normal and tumor microvessels
`
`is inversely correlated with the size of molecules
`
`Table l195456 presumably because of the size exclu
`sion effect of pores in the endothelium and the
`extracellular matrix surrounding endothelial cells
`The vascular permeability in tumors193738 however
`is less sensitive to the molecular weight compared
`with that in normal tissues343639 The reduced sensi
`tivity in tumors is probably due to large pores in
`tumor vessels as discussed earlier 194348 because
`the
`permeability is susceptible to the molecular size only
`the size is comparable to the dimension of pores in
`the vessel wall
`
`if
`
`5
`
`In addition to the steric effect the vascular barrier
`is selectively permeable to charged molecules5386062
`the barrier is more permeable to cationic
`In general
`or neutral molecules than anionic ones 5386062
`pre
`sumably owing to the negative charge of the base
`ment membrane and the extracellular matrix layer
`on the luminal surface of the vessel
`glycocalyx
`Wa11386365 For instance the vascular permeability to
`ribonuclease net charge +4 MW 13683 in the
`frog mesentery is approximately twice as high as that
`to otlactalbumin a molecule with similar size MVV
`14176 but negative charge net charge
`1061 The
`same trend has been observed in tumors62 Mecha
`nisms of the charge selectivity are still controversial
`Electron microscopy studies have demonstrated that
`cationized ferritin binds to glycocalyx and basement
`membrane in normal vessels5606466
`suggesting that
`the charge effect on the vascular permeability is
`
`

`

`168
`
`Fan Yuan
`
`mediated through electrostatic binding or repulsion
`between tracer molecules and the vessel wall Smit
`and Comper63 however proposed that electrostatic
`interactions between albumin and other polyions
`conditions The
`were negligible under physiological
`is an important
`charge selectivity of the vessel wall
`issue in gene delivery Although the cationic charge
`of delivery vehicles eg polycationic
`liposomes67 and
`amino polymers68 may improve the efficiency of
`gene transfer into cells the electrostatic binding of
`to the vessel wall may significantly
`the vehicles
`influence the pharmacokinetics
`of gene delivery
`
`Convection Versus Diffusion
`
`The previously mentioned study of convection
`versus diffusion provides only qualitative results be
`cause the single vessel perfusion technique itself can
`cause an increase in tumor vascular permeability to
`macromolecules21 The same problem has also been
`tissue isolated tu
`encountered
`in the study of
`mors7374 in which the fluid loss from the periphery of
`some tumors perfused ex vivo was approximately one
`order of magnitude higher than that from nonper
`fused tumors in vivo The perfusion induced vascular
`ingredients of
`leakiness may be caused by chemical
`has been shown that
`the
`the perfusate because
`permeability of vessels perfused with albumin solu
`that perfused with serum54
`than
`tion is higher
`Although the exact mechanism remains to be deter
`mined3436 the perfusion induced vascular
`leakiness
`can be exploited for improving drug delivery to solid
`tumors during the isolated perfusion of the limb7576
`the kidney77 the lung78 and the liver79
`
`it
`
`Effect of the Organ Microenvironment
`on the Vascular Permeability
`
`The vascular permeability of tumors may depend on
`the tissue microenvironment Our previous studies
`demonstrated that
`the vascular permeability of a
`human colon adenocarcinoma
`transplanted in cra
`nial windows or in the liver was higher than the
`the same tumor
`permeability of
`transplanted in
`dorsal skinfold chambers192023 Similarly growth fac
`torinduced vessels in collagen gels were more leaky
`to macromolecules when gels were transplanted in
`mouse cranial windows in comparison with the same
`cham
`transplanted in mouse dorsal skinfold
`gel
`bers80 Therefore how is the microvascular permeabil
`ity determined in vivo
`Stewart and Wileys demonstrated based on the
`study of quail chick transplantation chimeras that
`newly formed vessels in brain grafts transplanted in
`tissues were similar to the BBB In
`the abdominal
`contrast mesodermal grafts
`transplanted in the
`the BBB81 In another study
`brain did not possess
`Vajkoczy et al82 found that vessels in rat pancreatic
`islets transplanted in hamster dorsal skinfold cham
`bers were structurally similar to those in normal rat
`islets containing diaphragmed feriestrae
`pancreatic
`although they were originated from nonfenestrated
`vessels Both studies de
`hamster subcutaneous
`scribed above suggest that endothelial microenviron
`ment instead of the origin of vessels determines the
`vascular structure Our preliminary study of brain
`suggested that host environment
`tumors however
`
`is
`
`involves both
`vessel wall
`Transport of drugs across
`Diffitsion is the random
`diffusion and convection
`motion of molecules or small particles The mass flux
`is from high concentration to low concentration re
`gions and is proportional
`to the concentration differ
`ence between these two regions Convection is medi
`of fluid The fluid flux is
`ated by the movement
`determined by the balance between the hydrostatic
`and osmotic pressures In normal tissues convection
`the dominant mode of transport
`for macromol
`ecules whereas diffusion is more important
`for small
`molecules69 The situation in tumors can be signifi
`cantly different however
`because of the vascular
`leakiness and the interstitial hypertension To under
`stand mechanisms
`that govern the transvascular
`transport in tumors Lichtenbeld et a121 quantified
`the effective microvascular permeability in a human
`colon adenocarcinoma LS174T transplanted in the
`mouse dorsal skinfold chamber using the single
`vessel perfusion technique They found that the vas
`cular permeability to albumin was independent of
`the perfusion pressure in the range of 20 to 35 mm
`Hg indicating that convection was not
`the dominant
`mode of transport across
`the vessel wall The study
`also implies that diffusion is the dominant mode of
`transport in nonperipheral regions in solid tumors
`where the microvascular
`
`to
`
`pressure is nearly equal
`fluid pressure7° Convection may play
`the interstitial
`a role in drug delivery to peripheral tumor tissues
`however where a significant drop of the interstitial
`fluid pressure occurs71 This pressure gradient facili
`in the periph
`tates extravasation of macromolecules
`ery and causes extravasated macromolecules oozing
`from the tumor71 Consequently higher accumula
`tion of macromolecules eg monoclonal antibodies
`may be observed at the interface between tumor and
`normal tissues72
`
`

`

`Transvascular Drug Deliveg in Solid Tumors
`
`169
`
`might also play a role in the regulation of the vascular
`structure For example the vascular permeability of
`a human glioblastoma transplanted in the cranial
`window was comparable to that of the BBB The tight
`B IB disappeared
`however when the same tumor
`was transplanted subcutaneously72 In addition the
`cutoff size of pores in vessels of tumors transplanted
`in cranial windows was smaller than that in the same
`tumor line but transplanted in dorsal skinfold cham
`bers43 Therefore our hypothesis is that diffusible
`released by host eg skin brain and
`cytokines
`transplanted eg brain tumors cells as well as
`cells eg macrophages
`inflammatory
`fibroblasts
`penetrated into the grafts interact with each other in
`the microenvironment of endothelial cells and the
`balance of interactions determines the vascular per
`meability
`Putative cytokines which may affect
`the vascular
`positive factors eg VEGF
`include
`permeability
`VPF that increase the microvascular permeability
`inhibitors of stimuli and regulators that cause forma
`tion of tight endothelial junctions Identification of
`specific cytokines which play a dominant role in a
`given scenario remains an important area of re
`search Furthermore some cytokines are multifunc
`tional VEGFVPF is a survival
`factor
`for newly
`formed vessels83
`in addition to the stimulation of
`vascular leakiness and angiogenesis84 Thus depriva
`tion of VEGFVPF induces vessel
`regression in both
`tumors20 and the retina of the premature newborn83
`
`Modulation of the Vascular Permeability
`
`Chemical modulation Dexamethasone
`a potent
`synthetic glucocorticoid and other steroids are able
`to reduce the permeability of normal and tumor
`vessels8687 The efficacy of dexamethasone to reduce
`edema caused
`by brain tumors or brain
`cerebral
`surgery is related to this effect1485
`In a study
`dexamethasone was used to reduce the tumor inter
`
`fluid pressure via the reduction in the vascular
`stitial
`permeability88 Pretreatment of tumors with dexa
`methasone however may also decrease
`the delivery
`therapeutic agents to brain and subcutaneous
`of
`tumors86 The action of dexamethasone may be
`mediated through the down regulation of VEGF
`VPF expression in tumor cells as well as the vascular
`response to permeability factors8789
`Vasoactive agents eg leukotrienes bradykinin
`and histamine have been used to enhance the
`monoclonal antibody uptake in solid tumors90 Vaso
`active agents are especially useful
`for the selective
`
`enhancement of drug delivery to brain tumors be
`cause the normal BBB is protected by a biochemical
`that blocks the effect of these agents91 The
`barrier
`in tumors is either absent or
`biochemical barrier
`incomplete91 Therefore infusion of vasoactive agents
`or vasoactive
`immunoconjugates may selectively open
`the BIB9°91 Ackerman et a192 pointed out however
`that vessels in rat Walker carcinosarcomas were less
`responsive to vasoactive agents than normal vessels
`in the liver surrounding tumors suggesting that the
`efficacy of vasoactive
`agents depends on the local
`biochemical environment and the expression of appro
`priate receptors91
`Among all permeability factors studied
`so far
`VEGFVPF is one of the most potent ones It can
`increase the vascular permeability at a concentration
`of less than 1 nM which is about 50000 times lower
`than the effective
`concentration
`of histamine24
`VEGFVPF is expressed in various tumor and normal
`cells eg fibroblasts macrophages epidermal keratin
`and the expression can be stimulated
`ocytes24849394
`by hypoxia differentiation and growth factors 24849395
`The overexpression of VEGFVPF is considered as
`one of the key mechanisms of the hyperpermeability
`of tumor vessels52024 because 1 neutralizing VEGF
`VPF in tumors significantly reduces the microvascu
`lar permeability20 and 2 the vascular permeability
`is in most cases correlated with the level of VEGF
`VPF expression in tumors transplanted in the same
`location2096 VEGFVPF may not be the only factor
`that determines the vascular permeability The per
`meability of a human colon adenocarcinoma
`trans
`planted in the liver was twice as high as that of the
`same tumor transplanted in the subcutaneous tissue
`although the VEGFVPF expression was significantly
`tumors than in subcutaneous tumors23
`lower in liver
`Furthermore local treatment of normal tissues with
`exogenous VEGFVPF causes only a transient
`in
`crease in the vascular permeability24 The persistence
`leakiness is less than 30 min
`time of the vascular
`utes24 which is qualitatively similar to the effects of
`agents eg histamine97 There
`other vasoactive
`in vascular endothelial cells must
`fore other changes
`be involved in maintaining the hyperpermeability of
`tumor vessels
`The signal
`transduction
`of the stimulation with
`VEGFVPF9899 or histarnine9899 involves nitric oxide
`NO synthesis in endothelial cells Hence inhibition
`of the NO synthesis may abolish the effect of these
`agents on the vascular permeability98 However
`NO synthase inhibitors cannot
`reduce tumor vascu
`
`

`

`170
`
`Fan Yuan
`
`lar permeability to the same level as normal vessels22
`suggesting the existence of additional mechanisms
`involved in the regulation of tumor vascular perme
`
`ability
`The modulation of the vascular permeability in
`normal
`tissues is an equally important
`issue as
`in tumors because tumor
`compared with that
`mediated or therapeutic agentmediated up regula
`tion of the normal vascular permeability may cause
`problems For example vaso
`severe
`pathological
`genic cerebral edema remains a common problem
`for many brain tumor patients there is a significant
`correlation between the edema and mortality14 The
`edema is caused partially by the disruption of the
`BBB and the treatment of the syndrome includes
`reduction in the vascular permeability with cortico
`steroids eg dexamethasone and nonsteroidal anti
`inflammatory agents eg ibuprofen485 Vascular
`leak syndrome VLS a consequence of immuno
`toxin and cytokine therapiesm1°3 is an example of
`side effects related to the increase in normal vascular
`permeability VLS is characterized
`by hypoalbumin
`emia peripheral edema and fluid retention in the
`is one of the limiting factors for improving
`body It
`cancer treatment with these agents Mechanisms of
`VLS are agent dependent
`It has been hypothesized
`VLS is caused
`by the
`that
`interleukin2induced
`activation of complement and contact systems as well
`as the leukocyte mediated endothelial cell injury101105
`VLS is the conse
`whereas immunotoxininduced
`quence of the disruption of endothelial cellextracel
`lular matrix interactionsm° Therefore VLS can be
`reduced to some extent by inhibitors of 1 comple
`ment and contact systems eg Clesterase inhibi
`tor1°1 2 leukocyte endothelial
`interactions eg
`dextran sulfate1°3 or 3 vasoactive
`agents eg
`dexamethasone 03
`Physical modulation Hyperosmolar solutions have
`been used to open the interendothelial junctions for
`improving drug delivery to brain tumors33107108 Both
`convection and diffusion can be increased dramati
`infusion of hyperosmolar
`cally after intravascular
`solutions The percent
`increase in drug exposure of
`is significantly less than that
`brain tumors however
`of normal brain tissues Consequently this type of
`treatment may negatively
`the therapeutic
`index of drugs35108
`Ionizing radiationio91 and hyperthermia2115 can
`increase the vascular permeability in both normal
`and tumor tissues These modalities when combined
`with systemic therapies may improve the efficacy of
`drugs via enhanced delivery
`
`affect
`
`Effect of the Transvascular Transport
`on Tumor Pathophysiology
`
`of
`
`The transvascular
`
`transport in tumors may affect the
`tumor pathophysiology
`In addition to the induction
`the vasogenic brain edema as mentioned ear
`lier1483 the transvascular
`transport may affect angio
`genesis tumor blood flow and interstitital hyperten
`sion
`
`Angiogenesis and Vascular Permeability
`Vascular leakiness has been suggested to be neces
`sary for angiogenesis to proceed24 The coexistence of
`is likely caused by three mecha
`these two processes
`nisms First the hyperpermeability of tumor vessels
`promotes angiogenesis through the enhanced extrava
`sation of plasma proteins24 The extravasated pro
`teins form cross linked fibrin and activate metallopro
`teases for digesting other extracellular matrix
`molecules These processes
`facilitate the migration
`of endothelial cells and fibroblasts which are neces
`sary for the formation of new blood vessels and other
`tumor stroma tissues24 Second VEGFVPF is
`a
`potent angiogenic factor and a potent permeability
`factor249394 Therefore angiogenic vessels are likely
`if they are induced by VEGFVPF
`hyperpermeable
`junctions between mitotic and
`Third endothelial
`dying cells are leaky to macromolecules 116117 Hence
`vessels containing mitotic cells may be
`these
`hyperpermeable
`to plasma proteins even
`if
`
`angiogenic
`
`vessels are induced by other angiogenic factors eg
`basic fibroblast growth factor Not all angiogenic
`formed sprouts
`vessels are leaky however Newly
`have been shown to be
`impermeable to carbon
`particles or liposomes of 90 nm in diameter Fig
`particles were localized
`11215 Most extravasated
`near the roots of capillary sprouts or in the perivascu
`lar regions of mature vessels Fig 11243 Further
`in a human glioblastoma
`more the vessel wall
`xenograft has been shown to be impermeable to
`small tracer molecules although the angiogenesis in
`this tumor model was comparable with others13
`The inconsistency
`discussed here suggests that
`the concomitance of angiogenesis and vascular leaki
`ness may depend on how angiogenesis is induced and
`which angiogenic factors are involved Kumar et ali18
`the effect of various angiogenic
`compared
`basic fibroblast growth factor VEGFVPF platelet
`factor platelet derived endothelial
`derived growth
`cell growth factor hepatocyte growth factor and
`interleukin8 on the permeability of endothelial cell
`monolayers in vitro and found that only VEGFVPF
`
`factors
`
`

`

`Transvascular Drug Deliveg in Solid Tumors
`
`171
`
`Future Studies
`
`caused a significant increase in the permeability The
`differences among various angiogenic factors have
`also been compared in terms of the stimulation of
`endothelial cell proliferation the inhibition of anchor
`age disruptiondependent apoptosis and the induc
`tion of endothelial cell adhesion and spreading on
`substratesn9 These results suggest
`that angiogenic
`factors other than VEGFVPF do not cause signifi
`cant vascular leakiness24118
`
`interstitial
`
`Tumor Blood Flow and Vascular Permeability
`The vascular permeability may influence tumor blood
`flow Netti et aim proposed
`based on computer
`simulations that the increase in the vascular perme
`between blood flow and
`ability allowed coupling
`fluid movement The coupled flow may
`diminish the blood pressure gradient along tumor
`vessels Thus the driving force for the blood flow can
`vessels
`be significantly reduced in hyperpermeable
`The coupling hypothesis suggests a new mechanism
`in addition to the elevated geometrical and viscous
`resistance in the tumor microcirculation6 for explain
`ing the heterogeneous blood flow and hypoxia in solid
`tumors Therefore down regulation of the microvas
`cular permeability may potentially improve tumor
`blood flow and oxygen supply which in turn en
`hances
`the efficacy of radiation therapy
`
`Interstitial Fluid Pressure and Vascular
`
`Permeability
`
`The interstitial
`fluid pressure is elevated in both
`animal and human tumors compared with the pres
`sure in no