A Non-Equilibrium Thermodynamic Model For Tumor Extracellular Matrix With Enzymatic Degradation

Journal of the Mechanics and Physics of Solids 104 (2017) 32–56
Contents lists available at ScienceDirect
Journal of the Mechanics and Physics of Solids

journal homepage: www.elsevier.com/locate/jmps
A non-equilibrium thermodynamic model for tumor

extracellular matrix with enzymatic degradation
Shi-Lei Xue a, Bo Li a, Xi-Qiao Feng a,∗, Huajian Gao b
a
Institute of Biomechanics and Medical Engineering, AML, Department of Engineering Mechanics, Tsinghua University, Beijing 100084, PR
China
b
School of Engineering, Brown University, Providence, RI 02912, USA
a r t i c l e i n f o a b s t r a c t
Article history: The extracellular matrix (ECM) of a solid tumor not only affords scaffolding to support tu-
Received 23 January 2017 mor architecture and integrity but also plays an essential role in tumor growth, invasion,
Revised 18 March 2017
metastasis, and therapeutics. In this paper, a non-equilibrium thermodynamic theory is es-
Accepted 5 April 2017
tablished to study the chemo-mechanical behaviors of tumor ECM, which is modeled as a
Available online 6 April 2017
poroelastic polyelectrolyte consisting of a collagen network and proteoglycans. By using
Keywords: the principle of maximum energy dissipation rate, we deduce a set of governing equations
Tumor for drug transport and mechanosensitive enzymatic degradation in ECM. The results re-
Mechanobiology veal that osmosis is primarily responsible for the compression resistance of ECM. It is sug-
Extracellular matrix gested that a well-designed ECM degradation can effectively modify the tumor microen-
Chemo-mechanical coupling vironment for improved efficiency of cancer therapy. The theoretical predictions show a
Transport good agreement with relevant experimental observations. This study aimed to deepen our
Enzymatic degradation understanding of tumor ECM may be conducive to novel anticancer strategies.
© 2017 Elsevier Ltd. All rights reserved.
1. Introduction
A solid tumor consists of cancer cells, host cells, and an extracellular matrix (ECM). The ECM not only provides a
mechanical scaffold regulating the behaviors of all cells in the tumor but also plays a crucial role in the development,
invasion, and metastasis of cancer. The physiological and pathological significance of tumor ECMs has been illustrated by a
wide diversity of clinical syndromes (Frantz et al., 2010; Lu et al., 2012). In spite of their tissue-specific nature, the ECM can
be considered as a polyelectrolyte gel primarily consisting of a collagen fibril network entangled with glycosaminoglycans
(GAGs) (Fig. 1), which carry negative charges and imbibe water (Aukland, 1981; Levick, 1987; Swartz and Fleury, 2007;
Provenzano et al., 2012; Stylianopoulos et al., 2012). Collagen fibers, constituting up to 90% of the ECM volume, are
remarkably stiff in tension and endow the tumor tissues with a higher tensile strength (Frantz et al., 2010; Gilkes et al.,
2014; Jain et al., 2014). By contrast, GAGs are responsible for the compressive resistance of tumors owing to their capacity
to trap water (Frantz et al., 2010; DuFort et al., 2016). This gel-like structure confers elasticity, porosity, insolubility, and
other physical properties on ECM, and determines its role in scaffolding to sustain mechanical stresses and transport fluids.
Since anticancer drugs need to penetrate through the vessel wall and ECM to reach the target cancer cells, the mechanical
and transport properties of the ECM play a vital role in drug-based cancer treatment (Jain and Stylianopoulos, 2010; Miao
et al., 2015). However, the differential growth of tumors commonly elicits compressive stresses in the surrounding ECM as a
∗
Corresponding author.
E-mail address: fengxq@tsinghua.edu.cn (X.-Q. Feng).
http://dx.doi.org/10.1016/j.jmps.2017.04.002
0022-5096/© 2017 Elsevier Ltd. All rights reserved.
S.-L. Xue et al. / Journal of the Mechanics and Physics of Solids 104 (2017) 32–56 33
Nomenclature
Force and deformation

P First Piola–Kirchhoff stress tensor
σ (σ i ) Cauchy stress tensor (and its principal components)
σc (σ c ) Applied uniaxial compressive stress (and its absolute value)
Overall fluid pressure
Osmotic pressure
Electric potential
f Force along collagen fibril
p Fluid pressure in external solution
F(λi ) Deformation gradient tensor (and its principal components)
C Right Cauchy–Green tensor
J Volumetric ratio of deformation
Energy
E Internal energy of ECM per unit volume in reference configuration
hα Enthalpy per fluid particle of constituent α
η Entropy per unit volume in reference configuration
ηα Entropy per fluid particle of constituent α
W Helmholtz free energy per unit volume in reference configuration
μα Gibbs free energy (i.e. chemical potential) per fluid particle of constituent α
ς Energy dissipation rate per unit volume in reference configuration
Physical property
Cα , cα Nominal concentrations of fluid constituent α in reference and current configurations
Q α , qα Mass fluxes of fluid constituent α
Vα , vα Relative velocities of fluid constituents
hα s , hαβ Diffusive drag coefficients of fluid constituents
ξα Generation rate of fluid constituent α
ϕ m , ϕˆ m Volume fractions of solid component m in reference and current configurations
ν α , ν m Volumes of a fluid particle of constituent α and a solid fibril (chain)
rα , rm Radii of a spherical fluid particle of constituent α and cross-section of a solid fibril (chain)
zα , z2 Electric charges of a fluid particle of constituent α and a GAG chain
dm Degradation factor pertaining to solid component m
kem Enzymatic degradation rate
γm Mechanosensitive factor
E1 Tensile elastic modulus of collagen fibril
Da , Da0 Diffusivities in the ECM and free solution
k Permeability of the ECM
ηvis Viscosity of water
χ Dimensionless measure of enthalpy of mixing
kB Boltzmann constant
T Temperature
result of the rapid proliferation of cancer cells. High compressive stresses can convert fibroblasts into contractile myofibrob-
lasts that are capable of producing new ECM, which further increases the stresses in and stiffens the tumor. The accumulated
stresses and the enhanced stiffness may compress the blood vessels embedded in the ECM, impeding perfusion and delivery
of anticancer drugs (Jain et al., 2014). In addition, experimental measurements and histological examination have revealed
that the ECM in solid tumors is distinctly different from that in normal tissues. Generally, tumor interstitium is featured by
higher collagen contents, and a lower diffusive coefficient (Jain, 1987). The dense ECM tends to block the diffusion of thera-
peutic particles (e.g., nanomedicines) and confine them in the vicinity of the blood vessels (Chauhan and Jain, 2013). These
facts suggest that the tumor ECM can pose a physical barrier for the delivery of anticancer agents such as nanomedicines
containing macromolecular drug particles. Therefore, the normalization or degradation of tumor ECM has been suggested as
an effective strategy to improve the efficacy of therapeutic drugs (Jain and Stylianopoulos, 2010; Jain, 2013). Degrading the
ECM components can help alleviate stress levels, improve tumor perfusion, and increase the accessible volume for accom-
modating mobile drug particles. To this end, enzyme molecules, hormone relaxin, transforming growth factor-β blockade,
and tumor penetrating peptide have been shown to facilitate delivery of nanotherapeutics in solid tumors by enhancing
interstitial transport (Mckee et al., 2006; Mok et al., 2007; Perentes et al., 2009; Liu et al., 2012; Sugahara et al., 2010).
34 S.-L. Xue et al. / Journal of the Mechanics and Physics of Solids 104 (2017) 32–56
Fig. 1. Schematic of ECM in a solid tumor.
Due to the significance of ECM in cancer development and therapeutics, it is of great interest to understand its me-
chanical behavior under different biological and therapeutic conditions. In particular, a better understanding of the matrix
degradation strategy will require a quantitative evaluation on how the mechanical stresses and transport property of the
tumor ECM affect the efficacy of drug-based therapeutics. In the present paper, a thermodynamic framework integrating
the principles of mechanics and chemistry is formulated to describe the mechanical behaviors and degradation mechanisms
of tumor ECMs. The transport and degradation equations are derived in terms of the maximum dissipation rate principle.
A specific poroelastic theory accounting for mechanical, chemical, and biological cues at the microscale is proposed to
study nonlinear deformation, drug transport and degradation of the ECM. A stress-dependent enzyme reaction equation is
deduced to characterize the enzymatic degradation process. This model might be useful in designing innovative anticancer
drugs and strategies combining mechanotherapy and chemotherapy of tumors.
The organization of this paper is as follows. Section 2 presents a general thermodynamic framework of poroelastic solids
with degradation effects, somewhat analogous to continuum damage mechanics. In Section 3, this framework is specified
to describe the properties and enzymatic degradation of tumor ECM. In Section 4, we use the proposed model to illustrate
the compression resistance of ECM and the effects of mechanical modulation and enzymatic degradation in drug delivery.
Finally, Section 5 summarizes the main conclusions of this study.
2. Theoretical framework of poroelastic solids
We treat the tumor ECM as a poroelastic polyelectrolyte gel capable of transporting fluids. The collagen fibril network
and GAGs are considered as a solid skeleton (Fig. 1). The ECM space is filled with fluids consisting of solvents (usually
water) and solutes (e.g., ions and drugs). Gibbs (1878) and Biot (1941, 1972) formulated a general thermodynamic framework
for a porous solid. This poroelastic theory provides a general approach to formulating continuum mechanics theories for
deformable porous media (Coussy, 2004; Xue et al., 2016) and allows for mathematical representations of the complex
coupling among solid and fluid elements (Cowin and Cardoso, 2012). On the basis of the traditional poroelasticity, here
we develop a thermodynamic framework to describe the deformation, transport and degradation of a tumor ECM from the
continuum perspective of homogenized microstructures.
Let X and x denote the positions of a representative material point in the initial (reference) configuration and the
current configuration, respectively, F = ∂ x/∂ X being the associated deformation gradient tensor. For a poroelastic solid, the
balance equation of a fluid constituent α can be expressed as (e.g., Hong et al., 2008)
C˙ α = −∇R · Q α + ξα , (1)
where Cα is the nominal concentration, Q α the nominal flux, and ξ α the reaction-induced generation rate of the constituent
α.
Let V and S be the volume and surface area occupied by the porous medium in the reference configuration, respectively.
The balance of internal energy in the poroelastic solid can be expressed as

d
EdV = P : F˙ dV + hα ξα dV − hα Q α · dS, (2)
dt V V α V α ∂V
where E is the internal energy per unit volume in the reference configuration, P the first Piola–Kirchhoff stress, and hα the
enthalpy per fluid particle of the constituent α .
The Clausius–Duhem inequality can be written as

d
ηdV ≥ ηα ξα dV − ηα Q α · dS, (3)
dt V α V α ∂V
where η is the total entropy and ηα is the entropy per fluid particle of the constituent α .
Let W = E − ηT denote the Helmholtz free energy in the whole system, and μα = hα − ηα T denote the Gibbs free energy
of the fluid constituent α , where T is the absolute temperature. According to Eqs. (2) and (3), we have

d
W dV ≤ P : F˙ dV + μα ξα dV − μα Q α · dS. (4)
dt V V α V α ∂V
Its differential form is

W˙ ≤ P : F˙ + μα C˙ α − ∇ R μα · Q α . (5)
α α
A poroelastic solid may contain multiple solid and fluid components, and can also be subject to chemical degradation.
In such cases, the Helmholtz free energy of the system can be expressed as W = W (F, Cα , dm ), where dm stands for the
degradation factor pertaining to the solid component m. The differential of W is written as
∂W ˙ ∂W ˙ ∂W
W˙ (F, Cα , dm ) = :F+ Cα + d˙ . (6)
∂F α
∂ Cα m
∂ dm m
Ziegler (1977) stated that the constitutive equations of a material can be determined by its Helmholtz free energy and
dissipation function. For the poroelastic solid, the dissipation rate can be expressed as

ς = P : F˙ + μα C˙ α − ∇R μα · Q α − W˙ . (7)
α α
From Eq. (5), it can be seen that ς ≥ 0 always holds.
Substituting Eq. (6) into (7) leads to

∂W ∂W ˙ ∂W
P− : F˙ + μα − Cα − ∇ R μα · Q α − d˙ = ς . (8)
∂F α
∂ Cα α m
∂ dm m
Considering the arbitrariness of F˙ and C˙ α , we obtain the following constitutive equations:
∂W
P= , (9)
∂F
∂W
μα = . (10)
∂ Cα
Using Eqs. (9) and (10), the dissipation Eq. (8) reduces to
∂W
− ∇ R μα · Q α − d˙ = ς . (11)
α m
∂ dm m
The two terms at the left-hand side of Eq. (11) correspond to the movement of the fluid constituents and the degradation
of the solid constituents, respectively.
The energy dissipation arises from two primary mechanisms. One is the hydrodynamic frictions between the moving
fluid constituents and the surrounding media in the poroelastic solid, and the other is the loss of energy that the leaving
solid mass takes away during degradation. Let ς V and ς d denote the rates of energy loss due to the two mechanisms,
respectively. The total rate of energy loss is written as

ς dm , d˙m , Vα = ςV (dm , Vα ) + ςd dm , d˙m , (12)
where Vα is the velocity of fluid constituent α relative to the solid, and d˙m is the degradation rate of solid component m.
Note that ς d and ς V are coupled and vary with the degradation factor dm , since the dissolution of solid components may
affect the concentrations of fluid constituents. Combining Eqs. (11) and (12) and using the relation Q α = Cα Vα , we obtain

ςV = − Cα ∇R μα · Vα , (13)
α
∂W
ςd = − d˙ . (14)
m
∂ dm m
Following Ziegler (1977) and other researchers (Rajagopal and Srinivasa, 2004; Rajagopal et al., 2007; Abu Al-Rub and
Darabi, 2012; Pan and Zhong, 2014), we utilize the principle of maximum dissipation rate to deduce the constitutive model.
According to this principle, a physical process tends to happen in the direction that maximizes the energy dissipation rate.
Eqs. (13) and (14) can be regarded as the constraints of energy dissipation processes. Therefore, the principle of
maximum dissipation rate turns out to maximize the following auxiliary function

∂W
y = ς − L1 Cα ∇R μα · Vα + ςV − L2 d˙ + ςd , (15)
α m
∂ dm m
where L1 and L2 are the Lagrange multipliers. The extrema of y can be obtained by setting its first derivatives with respect
to Vα and d˙m to be zero, which leads to
∂y ∂ ςV
= ( 1 − L1 ) − L1Cα ∇R μα = 0, (16)
∂ Vα ∂ Vα
∂y ∂ς ∂W
= ( 1 − L2 ) d − L2 = 0. (17)
∂ dm
˙ ∂ dm
˙ ∂ dm
Setting u1 = −(1 − L1 )/L1 , u2 = −(1 − L2 )/L2 , we have
∂ ςV
−Cα ∇R μα = u1 , (18)
∂ Vα
∂W ∂ ςd
− = u2 . (19)
∂ dm ∂ d˙m
Substituting Eqs. (18) and (19) into Eqs. (13) and (14) leads to
−1 −1
∂ ςV ∂ ςd ˙
u1 = ςV ·V , u2 = ςd dm . (20)
α
∂ Vα α m ∂ d˙m
Eqs. (18) and (19) are the evolution laws governing the movement of the fluid constituents and the degradation of
the solid components in the poroelastic solid, and they will be used in Section 3 to deduce the transport equations and
degradation law, respectively.
3. Tumor ECM model
As a poroelastic solid, the tumor ECM can sustain both mechanical stresses and transport fluids, which play important
roles in cancer development and therapeutics. In addition, the enzymatic degradation of tumor ECM can reduce the
resistance of fluid transport, lower the stress level in the tumor, and thus has been recognized as a promising anticancer
strategy. In the present study, therefore, we develop an integrated theoretical model accounting for the mechanical defor-
mation, fluid transport, and enzymatic degradation of ECM. In this Section, a tumor ECM model is presented on the basis
of the theoretical framework of poroelastic solids formulated in Section 2. We will first discuss the constitutive relations
of the ECM. Next, we will derive the transport equations and link transport parameters to the physical properties of the
ECM and the transported particles. Finally, the enzymatic degradation law of the ECM will be provided by considering the
molecular mechanisms of enzyme–ECM interaction.
3.1. Constitutive equations
Fibrillary collagen and GAGs are the most important components determining the mechanical and transport properties
of tumor ECM. In such ECM, collagen fibrils are woven together, forming a fibrous network that provides the structural
and mechanical integrity. Proteoglycans and hyaluronan are large brush-shaped macromolecules consisting mainly of
GAG chains, which interpenetrate into the collagen network without directly sustaining mechanical loads. Entangled and
immobilized by the collagen network, the GAGs can imbibe the solution and carry fixed negative charges via disaccharide
units (Aukland, 1981; Levick, 1987; Swartz and Fleury, 2007; Netti et al., 20 0 0).
Since the ECMs in solid tumors are basically polyelectrolyte gels capable of bearing charges and imbibing water, the
theory of polyelectrolytes (Hong et al., 2010) can provide insights into the macroscopic mechanics of tumor ECM. Here, a
theory of two-component polyelectrolyte gel is established to capture the basic features of tumor ECM. The undeformed
dry ECM is chosen as the reference state. Let ϕ 1 and ϕ 2 stand for the volume fractions of collagen fibrils and GAG chains in
the reference configuration. The porous nature of the stress-free collagen network implies ϕ1 + ϕ2 < 1. The Helmholtz free
energy density of the ECM contains three main parts, W = Wnet + Wsol + Wion , where Wnet is the elastic strain energy of the
collagen network, Wsol the free energy due to the mixing of the GAG chains and the solvent, and Wion the contribution due
to the mixing of the solvent and solutes. The expressions of these free energies will be derived below.
The eight-chain hyperelastic model proposed by Arruda and Boyce (1993) can describe the mechanical behavior of
porous solid skeletons (Palmer and Boyce, 2008; Kuhl et al., 2005). Here, we use this model to link the deformation of

the collagen network to the stretch of collagen fibrils. As the collagen network is deformed, the stretch ratio of a single
fibril can be expressed as λ = I1 /3, where I1 = tr(FT · F ). For simplicity, the elasticity of a single fibril is characterized by
f = E1 A1 (λ − 1 ), where f denotes the force along the fibril, and E1 and A1 are its elastic modulus and cross-sectional area,
respectively (Soulhat et al., 1999). Considering the different properties of a collagen fibril under tension and compression,
we set E1 = E1 for λ ≥ 1 and E1 = 0.1E1 for λ < 1, with E1 being the tensile Young’s modulus. The free energy of a single
l λ
collagen fibril is then w1 = f dl¯ = l0
l0 f (λ̄ )dλ̄ = 1 E v1 (λ − 1 )2 , where ν1 = A1 l0 is its volume. The free energy density of
1 2 1
the collagen network is Wnet = n1 w1 , where n1 is the number density of collagen fibrils. Thus Wnet can be written as
1
Wnet = E1 ϕ1 ( I1 /3 − 1 )2 , (21)
2
where ϕ1 = n1 v1 .
Making use of the Flory–Huggins theory (Flory, 1942; Huggins, 1942), the free energy Wsol resulting from the mixing of
the GAG chains and the solvent can be expressed as

νwCw χ ϕ2
Wsol = kB T Cw ln + , (22)
ϕ2 + νwCw ϕ2 + νwCw
where Cw is the concentration of the solvent in the reference configuration, ν w the volume of a water molecule, kB the
Boltzmann constant, T the absolute temperature, and χ a dimensionless parameter characterizing the enthalpy of mixing.
The contribution from mixing the fluid constituents is (Hong et al., 2010)

Cb
Wion = kB T Cb ln −1 , (23)
b
νwCw cbref
where Cb is the concentration of the solute b in the reference configuration and cbref the reference concentration of solute b.
In addition, we introduce two constraint conditions in the description of the tumor ECM. First, we assume that the solid
and fluid constituents of the ECM are incompressible in volume. The condition of incompressibility reads

ϕ1 + ϕ2 + νβ Cβ = J, (24)
β
where ν β is the volume per particle of species β , and J = det(F ) is the volumetric ratio. Second, the system is assumed to
be electroneutral everywhere. It is worth mentioning that there may exist an electric double layer near the interface of the
gel and the external solution (Marcombe et al., 2010). However, since the electric double layer (scaled by the Debye length)
is very thin, electroneutrality is assumed to prevail throughout the ECM (Gu et al., 1998; Marcombe et al., 2010; Cheng et
al., 2015). This condition can be expressed as
ϕ2
ezβ Cβ + ez2 = 0, (25)
β
v2
where e is the elementary charge, zβ the valence of fluid constituent β , z2 the electric charge carried by each GAG chain,
and ν 2 the volume per GAG chain. In Eq. (25), the first term corresponds to the total electric charges of all mobile particles,
while the second term denotes the fixed charges contributed from the GAG chains.
Combining Eqs. (24) and (25), the Helmholtz free energy density is derived as

ϕ2
W = (1 − d1 )Wnet + Wsol + Wion + ϕ1 + ϕ2 + νβ Cβ − J + ezβ Cβ + (1 − d2 )ez2 , (26)
β β
v2
where and are the Lagrange multipliers, which can be interpreted as the overall fluid pressure and electric potential,
respectively. To incorporate the degradation of ECM, we define degradation factors d1 and d2 to quantify the degradations
of the collagen fibrils and the GAG chains, respectively. The decay of the solid components due to enzymatic action is
modeled as a loss of volume fractions of collagen fibrils and GAG chains.
Substituting the Helmholtz free energy in Eq. (26) to Eq. (9) and using the Piola transformation, the Cauchy stress
σ = J−1 P · FT is identified as

1 3 1
σ = (1 − d1 )ϕ1 E1 1 − F · FT − I. (27)
3 I1 J
Using Eqs. (10) and (26), the electrochemical potentials of the solvent and solutes are expressed respectively as

∂W νwCw ϕ2 χ ϕ22 Cb
μw = = kB T ln + + − + νw , (28)
∂ Cw ϕ2 + νwCw ϕ2 + νwCw (ϕ2 + νwCw )2 b
Cw
∂W Ca
μa = = kB T ln + νa + eza . (29)
∂ Ca νwCw caref
3.2. Transport equations
Interstitial fluid flow and solute transport in biological tissues have attracted much attention (Swartz and Fleury,
2007). It has been revealed that interstitial fluid flow can affect cell–cell communication, morphogenesis, and cancer
metastasis, while the transports of proteins, nutrients, and drugs are basic processes involved in physiological activity,
signaling transduction, and therapeutics (Swartz and Fleury, 2007). In biological systems, mass transport involves multiple
mechanisms, e.g. diffusion and convection (Jain, 1987; Jain, 1990; Ramanujan et al., 2002). We will derive the generalized
transport equations by using the theoretical framework established above. Let qα = cα vα denote the flux in the current,
deformed configuration, where cα = J −1Cα and vα are the concentration and velocity of fluid constituent α in the current
configuration, respectively. According to Q α = Cα Vα and the Piola transformation qα = J −1 F · Q α , one has vα = F · Vα . In the
current state, Eqs. (13) and (18) become

− cα ∇ μα · vα = ςv (dm , vα ), (30)
α
∂ ςv
−cα ∇ μα = u1 , (31)
∂ vα
The hydrodynamics-induced dissipative rate per unit volume in the current configuration is ςv = ςV /J, and the parameter

u1 can be recast as u1 = ςv ( α ∂∂ vςv · vα )−1 .
α
The migrating fluid-carried particles interact with other particles and the surrounding solid components, leading to
energy dissipation. The transport dissipation can be expressed as (Rajagopal and Srinivasa, 2004)
1
ςv = hβγ (vβ − vγ )2 + hβ s vβ 2 , (32)
2
β γ β
where the first term stands for the interaction of fluid constituents, and the second term accounts for the interaction
between fluid constituents and solid components; hαβ and hα s are the drag coefficients related to the interaction among
fluid constituents and that between the fluids and the solid components, respectively. Combining Eqs. (31) and (32) yields

−cα ∇ μα = hαβ (vα − vβ ) + hα s vα . (33)
β
By distinguishing between solvent and solutes, Eq. (33) can be rewritten as

−ca ∇ μa = hab (va − vb )+haw (va − vw ) + has va , (34)
b

−cw ∇ μw = hwb (vw − vb ) + hws vw , (35)
b
where va and vw denote the relative velocity vectors of solute a and the solvent (i.e., water), respectively.
Eqs. (34) and (35) reveal that the chemical potential gradients of the solvent and the solutes serve as the driving forces
for fluid movement. These driving forces are balanced by the drag forces arising from the interactions between different
constituents in the tumor ECM. For simplicity, the solute-against-solute transport drag is ignored here (i.e., hab = 0) and,
thereby, the relative velocities of the solvent and the solutes are derived as

c w ∇ μw + hbw
b hbs +hbw cb ∇ μb
vw = − hbw hbs , (36)
hws + b hbw +hbs
1
va = (−ca ∇ μa + haw vw ). (37)
haw + has
The nonzero transport drag coefficients are given by (Ateshian et al., 2006)
1 kB T ca kB T ca
hws = , haw = , haw + has = , (38)
k Da0 Da
where k is the hydraulic permeability of the solvent in the ECM, Da0 and Da the solute diffusivity (i.e., diffusion coefficient)
in the free solvent and the ECM, respectively, and ca the solute concentration in the solvent.
Substituting Q α = Cα F−1 · vα into the fluid mass conservation Eq. (1), and making use of Eqs. (33)–(35), the transport
equations in the reference configuration are rewritten as

Db
C˙ w = ∇R · k˜Cw C −1
· Cw ∇R μw + C ∇ R μb , (39)
Db0 b
b

Da D Db
C˙ a = ∇R · Ca C −1
· ∇R μa + a k˜ Cw ∇R μw + C ∇ R μb +ξa , (40)
kB T Da0 Db0 b
b
where C = FT · F and k˜ is defined as

1 J D Cb

= + kB T 1 − b . (41)
˜k k Db0 Db0
b
In the tumor ECM, both collagen fibrils and GAG chains affect solute diffusion and solvent convection (Levick, 1987; Netti
et al., 20 0 0). The overall hindrance of solid components can be treated as a summation of the interactions between all solid
m
component and the mobile particles, that is, has = m hm as and hws =
m m
m hws . Here, has and hws are the drag coefficients
characterizing the interaction of solid components (such as collagen fibril and GAG chain) with the solutes and the solvent,
respectively. Similar to Eq. (38), they are expressed as
1 kB T ca
hm
ws = , haw + hm
as = , (42)
km Dam
where km is the hydraulic permeability of the meshwork of solid component m and Dam the diffusion coefficient of solute
a in the meshwork of polymer component m.
From Eqs. (38) and (42), the transport parameters of the ECM are related to those of solid components by
1 1 1 1 m−1
= , = − . (43)
k m
k m D a m
D am Da0
The Carmen–Kozeny equation has been widely used to link the permeability with specific matrix properties in tissues.
This equation is also employed here (Levick, 1987; Swartz and Fleury, 2007) to determine the expression of km , as given in
Appendix. Thereby and using Eq. (43), the permeability of the ECM can be written as
−1
1 8 ϕˆ m −1 1 − ϕˆ m
2
= − ηvis 2 ln ϕˆ m + ϕˆ m − 1 ϕˆ m − 3 + 2 ln ϕˆ m + , (44)
k 3 m
2
rm 1 + ϕˆ m
2
where ϕˆ m = (1 − dm )ϕm /J stands for the volume fraction of solid component m in the current configuration, rm is the chain
radius of solid component m, and ηvis is the viscosity.
We next consider the solute diffusion in the ECM. The free volume theory, the obstruction theory, and the hydrodynamic
theory have been proposed to characterize solute diffusion in porous solids (Amsden, 1998). The free volume theory
assumes that the solute diffuses within solvent spacing, rendering a redistribution of free volume. The obstruction theory
is based on the fact that the presence of impenetrable polymer chains causes an increase in the path length for diffusive
transport. The hydrodynamic theory considers the hydrodynamic frictions present in the solute diffusion (Amsden, 1998;
Masaro and Zhu, 1999). Among the three models, the obstruction model may provide the best prediction for rigid polymer
chains (Amsden, 1998), while the hydrodynamic model is suitable to describe the solute diffusion within the network of
flexible polymer chains (Cukier, 1984). Since the collagen fibrils are relatively stiff while the GAG chains are flexible, the
obstruction theory and the hydrodynamic theory are utilized here to model the diffusivities of collagen fibrils and GAG
chains, respectively (Amsden, 1998; Cukier, 1984). Thereby, we have
2
Da1 π ra /r1 + 1 Da2
= exp − , = exp −δ2 ra ϕˆ 20.75 , (45)
Da0 4 δϕ −0.5
1 ˆ1 +1 Da0
where ra denotes the radius of the solute particle, and δ 1 and δ 2 are two parameters. Da0 can be estimated by the
Stokes–Einstein formula Da0 = kB T /(6π ra ηvis ).
Combining Eqs. (45) and (43), the diffusivity of the tumor ECM can be described by
2 −1
Da π ra /r1 + 1
= exp + exp δ2 ra ϕˆ 2
0.75
−1 . (46)
Da0 4 δ1 ϕˆ 1−0.5 + 1
As can be seen from Eqs. (44) and (46), the transport properties of the ECM mainly depend on the volume fractions and
radii of collagen fibrils and GAG chains.
3.3. Degradation law
ECM degradation is a feasible treatment strategy to modify the tumor microenvironment and improve the efficiency
of cancer therapy. Enzymes, such as collagenase and metalloproteinases, are often used to degrade the ECM (Netti et al.,
20 0 0; Brown et al., 2003; McKee et al., 2006; Mok et al., 2007; Magzoub et al., 2008; Provenzano et al., 2012). In high
collagen-content tumors (e.g., Hsts26t sarcoma and mu89 melanoma xenografts), for example, degrading the collagen
matrix with bacterial collagenase treatment may cause a two-fold increase in the diffusion area of antibodies (e.g., igG) and
a three-fold increase in the diffusion area of herpes simplex virus (McKee et al., 2006), both of which are nanomedicine
targeting cancer cells. Matrix metalloproteinases targeting GAG chains can also improve the distribution of herpes simplex
virus (Mok et al., 2007). In the following, we will focus on the enzymatic degradation of the ECM and formulate the
degradation law on the basis of the proposed theoretical framework.
The first-order kinetic model, which is a simplified version of the classical Michaelis–Menton model (Brown et al., 2003;
Metters and Hubbell, 2005), is adopted to describe the enzymatic degradation. The enzymatic degradation of biopolymers is
a complex process. In the present study, we mainly consider the decrease in the volumetric fraction of biopolymers, rather
than the detailed geometric changes, during enzymatic degradation.
The enzymatic degradation of solid components in a tumor ECM is evaluated by the loss of their volume fractions. Thus
the enzymatic kinetics can be expressed as
kcat
(ϕˆ m )• = − m e
c ϕˆ m , (47)
Km m
where kcat e
m and Km are two enzymatic kinetic constants, cm is the concentration of the enzyme that degrades the solid com-
•
ponent m. Introducing the enzymatic degradation rate as kem = −(ϕˆ m ) /ϕˆ m , which can be further related to the degradation
factor by kem = d˙m /(1 − dm ), Eq. (47) shows that the degradation rate without mechanical modulation can be expressed as
kem0 = cm
e kcat /K .
m m
Experimental observation evidenced that mechanical tension can stabilize collagen against enzymatic degradation
(Ruberti and Hallab, 2005; Camp et al., 2011; Flynn et al., 2013). In what follows, a general enzymatic kinetics model
incorporating the mechanical modulation will be deduced on the basis of the theoretical framework established in Section
2. We further assume that the enzymatic degradations of different components are independent reaction processes. Hence,

the overall degradation dissipation rate can be written as ςd = m ςdm , where ςdm is the rate of dissipation resulting from
the degradation of component m. Thus, Eq. (14) can be rewritten as
∂W ˙
ςdm = − d . (48)
∂ dm m
Following the principle of maximum energy dissipation rate, we obtain a series of equations similar to Eq. (19):
∂W ∂ςdm
− = um , (49)
∂ dm 2
∂ d˙m
∂ ςd ˙ −1
where um = ςd ( d ) . Eq. (49) is actually equivalent to (48), and both of them can be viewed as the degradation law.
2 ∂ d˙m m
Thus the driving force for degradation can be defined as Dm = − ∂∂dW , which is the work-conjugate of the degradation factor
m
dm .
Now we turn to deriving the expression of the degradation-induced dissipative rate ςdm by accounting for molecule
mechanisms and mechanical modulation. Each collagen molecule has a triple helical structure, which can protect against
enzymatic breakdown. When enzymes bind with collagen, a necessary step before hydrolysis of the chains is the helix
unwinding at the cleavage site (Perumal et al., 2008; Chang et al., 2012). However, tensile stresses may stretch the collagen
alpha into a tighter apposition, limiting the access of enzymes to the collagen molecules. In this process, mechanical
energy is dissipated by pulling looped structures, which form around the enzyme cleavage site, back into the triple-helix
conformation (Chang et al., 2012; Flynn et al., 2013) through both collagen dissolution and conformational change, that is,
ςdm = Dm0 d˙m + Dm1 ln (kem0 /kem )d˙m . (50)
Here Dm0 represents the free energy of the lost solid mass and is assumed to be proportional to the driving force of
degradation Dm , that is, Dm0 = (1 − βm )Dm , with β m being a constant. As mentioned above, the enzymatic degradation
kinetics is also related to the conformational change of collagen molecules. The energy consumed by pulling the unwinding
collagen molecules can be characterized by the reaction rate, and can be expressed as Dm1 ln(kem0 /kem ), Dm1 being a constant.
Substituting Eq. (50) into (49), we obtain the enzymatic degradation law accounting for the effect of mechanical
modulation as
d˙m = kem0 exp (−γm Dm )(1 − dm ), (51)
where γm = βm /Dm1 is a mechanosensitivity factor.
Eq. (51) can be employed to describe the enzymatic degradation of collagen fibrils. For enzymatic degradation of GAG
chains, mechanical loads have not been reported to influence the enzymatic reaction. Thus, γ 2 is set to be zero in this study.
In this Section, as an application of the chemomechanical theory proposed in Section 2, we have provided a constitutive
description of tumor ECM and formulate the associated kinetic laws that control the dissipation behaviors in tumor ECM.
The mechanical and chemical fields are governed by the constitutive Eqs. (27)–(29), the transport Eqs. (39) and (40), and
the degradation law (51). Combined with the mechanical equilibrium equation, specific boundary conditions, and initial
conditions, we can solve for the field variables including stresses, fluid chemical potentials, solute concentrations, and
degradation factors. The fluid transport, such as solvent convection and solute diffusion, is determined by the transport Eqs.
(39) and (40), with transport parameters derived by considering the specific interactions of fluid particles with the ECM.
Importantly, a mechanosensitive enzymatic degradation law (51) has been included in our theoretical framework.
4. Examples and discussions
The tumor ECM may impede the delivery of anticancer therapeutics by compressing blood vessels and blocking the
transport of therapeutic agents (Jain et al., 2014; Chauhan and Jain, 2013). Recent experiments have shown that the
enzymatic degradation of tumor ECM can improve drug efficacy by reducing the mechanical resistance and enhancing mass
transport in the ECM (Jain and Stylianopoulos, 2010; Jain, 2013). The model established in Section 3 can reveal the effects
of enzymatic degradation on the properties of tumor ECM. In this Section, we will first analyze the compression resistance
of the ECM and the transport efficacy of anticancer drugs, and then discuss the effects of enzymatic degradation on these
properties. The typical parameters used in the numerical calculation are adopted from literature and listed in Table 1.
Both fluid constituents in the external solution and the ECM contain solvent molecules (i.e., water) and free ions. For
simplicity, we consider two species of free ions, which carry a positive charge and a negative charge, respectively. In
Sections 4.3 and 4.4, drug particles and enzyme molecules will also be taken into consideration.
Table 1
Parameters adopted in the numerical simulations.
Symbol Description Value Source
E1 Tensile elastic modulus of collagen fibril 79 kPa Stylianopoulos and Barocas, 2007
ν Volume of a water molecule 3 × 10−29 m3
ν2 Volume of a GAG chain 3.8 × 10−27 m3 Levick, 1987
z2 Charge of a GAG chain –4.0 Levick, 1987
kB Boltzmann constant 1.38 × 10−23 JK−1
T Absolute temperature 298 K
χ Dimensionless measure of the enthalpy of mixing 0.2 Estimated
c0ref Reference concentration of free ions 9 × 1025 m−3 Lai et al., 1991
ηvis Viscosity of water 8.9 × 10−4 Pa s Hong et al., 2008
r1 Radius of collagen fibril 12.5 × 10−9 m Levick, 1987
r2 Radius of GAG chain 0.55 × 10−9 m Levick, 1987
ke0 Enzymatic degradation rate 6 × 10−4 s−1 Estimated
−1
γ Mechanosensitive factor 0.1kPa Estimated
4.1. Free swelling of tumor ECM
The tumor ECM is always in a swelling state with fluid perfusion in vivo. In this Subsection, free swelling of a tumor
ECM is considered to illustrate its mechanical and chemical properties. In such a situation, the electrochemical equilibriums
of the fluid particles prevail throughout the matrix.
The stress and the stretch ratio in a free swelling ECM are isotropic and homogeneous. We denote them as σ 0 and λ0 ,
respectively. When the ECM equilibrates with the external solution, we have σ0 = −p, where p is the fluid pressure in the
external solution. From Eq. (27), we get
1
= ϕ1 E1 λ−1
0 − λ0
−2
+ p. (52)
3
The electrochemical potentials of water and solute a in the external solution can be written as (Ateshian et al., 2006)

c̄a
μex
w = vw p − kB T c̄b , μex
a = kB T ln , (53)
b
caref
where c̄b is the concentration of free ion specie b in the external solution; b can be either “+” or “−”, representing positive
and negative free ions, respectively.
The uniformity of the electrochemical potentials of fluid constituents in the ECM and the external solution leads to
μw = μexw and μa = μa . From Eqs. (28) and (29), and Ca are derived as
ex

kB T νwCw ϕ2 χ ϕ22
Cb
=− ln + + +kB T − c̄b + p, (54)
νw ϕ2 + νwCw ϕ2 + νwCw (ϕ2 + νwCw ) 2
b
νwCw

eza + νa
Ca = νwCw c̄a exp − . (55)
kB T
The overall fluid pressure can be expressed as = + p, where is the osmotic pressure. It is known from Eq.
(52) that = 13 ϕ1 E1 (λ−1
0
− λ−2
0
), which indicates that the osmosis of the ECM bears the contractility of the collagen
network. The osmotic pressure can be further divided into two parts: the polymeric osmotic pressure pol arising from the
mixing of GAG chains and the solvent, and the ionic osmotic pressure ion due to the imbalance of ions in the ECM and
the external solution. They are expressed respectively as

kB T νwCw ϕ2 χ ϕ22
pol = − ln + + , (56)
νw ϕ2 + νwCw ϕ2 + νwCw (ϕ2 + νwCw )2

Cb
ion = kB T − c̄b . (57)
b
νwCw
Since the volume of solutes is usually much smaller than that of the solvent (i.e., water) in tumor ECM, we ignore the
contribution of the solute in the incompressibility condition (24), i.e., νwCw ≈ J − ϕ1 − ϕ2 . Thus pol becomes

kB T ϕ2 ϕ2 χ ϕ22
pol = − ln 1 − + + . (58)
νw J − ϕ1 J − ϕ1 ( J − ϕ 1 )2
In the external solution, the concentrations of positive and negative free ions are both set as c0 , that is, c̄+ = c̄− = c0 .
Because the volume of each free ion is very small, neglecting ν a has little impact on the electrochemical potentials of free
ϕ
ions. From Eq. (55), we have C+C− = (νwCw c0 )2 , which combined with the electroneutrality C+ − C− + z2 v 2 = 0 leads to
2

1 ϕ2 ϕ2 2
C± = ∓z2 + z2 + 4(νwCw c0 ) .
2
(59)
2 ν2 ν2
Thus ion is rewritten as

z ϕ2 2
ion = kB T 2
+ 4 c0 − 2 c0 .
2 (60)
ν2 J − ϕ 1 − ϕ 2
From Eqs. (55) and (59), we can express the normalized electric potential as
⎧ −1 ⎫

C ⎨
ν J − ϕ − ϕ 2 ⎬
e 1 − 1
∗ = = ln = ln 1 + 2 −1 + 1+4
2 1 2
c02 . (61)
kB T 2 C+ 2 ⎩ z2 ϕ2 ⎭
Then the stretch ratio λ0 of the free swelling ECM is determined from

1 kB T ϕ2 ϕ2 χ ϕ22
ϕ1 E 1 λ−1
0 − λ0
−2
=− ln 1 − + + 2
3 νw λ30 − ϕ1 λ30 − ϕ1 λ30 − ϕ1
⎡ ⎤
2
z2 ϕ2
+ kB T ⎣ + 4c02 − 2c0 ⎦. (62)
ν2 λ30 − ϕ1 − ϕ2
Once λ0 has been determined, the osmotic pressures and the electric potential can be derived from Eqs. (58), (60), and
(61). The results clearly show that the physical properties of collagen fibrils (i.e., ϕ 1 and E1 ) and GAG chains (i.e., ϕ 2 , z2 ,
and ν 2 ), as well as the ion concentration in the external solution (i.e., c0 ), play important roles in the deformation, osmosis,
and electrochemistry of tumor ECM. For a free swelling ECM, the volume ratio J is much larger than ϕ 1 and ϕ 2 . Therefore,
the osmotic pressures and electric potential can be approximated as

kB T
z2
2
pol ≈ − ln 1 − ϕˆ 2 + ϕˆ 2 + χ ϕ , ion ≈ kB T
ˆ 22 ϕˆ 2 + 4c02 − 2c0 , (63)
νw ν2
⎧ −1 ⎫
⎨
ν c 2 ⎬
1
∗ ≈ ln 1 + 2 −1 + 1+4
2 0
. (64)
2 ⎩ z2 ϕˆ 2 ⎭
It is found from Eqs. (63) and (64) that the volume fraction of GAG chains in the current configuration, ϕˆ 2 , directly
regulates the osmotic pressures and electric potential.
The effects of the volume fractions of collagen and GAG chains on the free swelling of ECM are depicted in Fig. 2. Fig.
2(a) and (b) show that increasing collagen volume fraction ϕ 1 leads to a smaller stretch ratio λ0 , a higher electric potential,
and higher osmotic pressures. This is because the collagen network acts as a skeleton to maintain the structural integrity,
and the denser collagen fibrils tend to prevent the ECM from swelling and deforming. The collagen network ensures the
concentration of fixed charges in the GAG chains not to be diluted by the ECM swelling. Thus, more collagen fibrils will
result in a larger electric potential and a higher ionic osmotic pressure in the ECM, as shown in Fig. 2(a) and (b). Similarly,
since the current volume fraction of GAG chains, ϕˆ 2 , declines as the ECM gradually swells, a denser collagen network will
cause a larger ϕˆ 2 , which corresponds to a higher polymeric osmotic pressure, as illustrated in Fig. 2(b). The above analysis
shows that the overall osmotic pressure also increases with increasing ϕ 1 . Besides, the tensile elastic modulus of collagen
fibrils, E1 , is found to exhibit a similar behavior with respect to ϕ 1 .
In contrast to collagen fibrils, GAG chains can promote the swelling of tumor ECM. As shown in Fig. 2(c), increasing the
volume fraction of GAG results in a larger stretch ratio. Interestingly, there may exist extremums in the electric potential
and the osmotic pressures at a specific volume fraction of GAG, as shown in Fig. 2(c) and (d). This can be explained as
follows. It can be seen from Eqs. (63) and (64) that the electric potential and the osmotic pressure vary with the
amount of GAG in the hydrated ECM, ϕˆ 2 . The ECM swells as ϕ 2 increases. The volumetric swelling tends to dilute the
concentration of GAG chains. Our calculations show that as ϕ 2 increases, the current volume fraction ϕˆ 2 of GAG first
increases and then exhibits a drop. Noting that the electric potential is negative, the absolute electric potential and the
osmotic pressure show similar trends with respect to ϕ 2 , as shown in Fig. 2(c) and (d).
The charges carried by each single GAG chain may be different in the ECM (Levick, 1987). As shown in Fig. 2(e),
increasing the amount of negative charges in each GAG chain z2 will result in a larger swelling ratio and a higher electric
potential. Correspondingly, the free swelling of ECM will lead to a higher ionic osmotic pressure. However, a larger swelling
ratio will lower the current volume fraction of GAG, rendering a lower polymeric osmotic pressure, as shown in Fig. 2(f).
Fig. 2. Free swelling of tumor ECM. The effects of collagen volume fraction ϕ 1 on (a) the stretch ratio λ and the normalized electric potential ∗ , (b) the
polymeric osmotic pressure pol , the ionic osmotic pressure ion , and the overall osmotic pressure . The effects of GAG volume fraction ϕ 2 on (c) λ,
∗ , (d) pol , ion , and . The effects of the negative charges z2 per GAG chain on (e) λ, ∗ , (f) pol , ion , and . The effects of normalized free ion
concentration c0∗ on (g) λ, ∗ , (h) pol , ion , and .
Fig. 3. Two modes of ECM compression: (a) confined compression in a standard experiment of poroelastic solids; (b) unconfined compression induced by
expanding tumor cells.
Therefore, the overall osmotic pressure, including the ionic osmotic pressure and the polymeric osmotic pressure, only
exhibits a modest variation with the amount of negative charges in each GAG chain. Comparing Fig. 2(d) and (f) shows that
ϕ 2 and z2 play different roles in regulating the osmotic pressures in the ECM.
In addition, the swelling behavior of ECM is also affected by the concentration of free ions. The free ions in the solution
tend to shield the effect of the fixed charges in the ECM. Therefore, increasing the amount of free ions will suppress
the ECM swelling and decrease the electric potential, as illustrated in Fig. 2(g). The normalized free ion concentration
c0∗ = c0 /c0ref , where c0ref = 9 × 1025 m−3 is the physiological value. Meanwhile, the weaker effect of fixed charges will
alleviate the ionic osmotic pressure ion , as shown in Fig. 2(h). However, since the polymeric osmotic pressure pol can
be elevated by ion concentration, the overall osmotic pressure, = ion + pol , only exhibits a weak dependence on the
concentration of free ions.
From the above discussions, we find that the deformation, electric potential, and osmotic pressure in a freely swelling
ECM are affected by its physiochemical properties and the external solutions in a complex way, and the factors of mechanics
and electrochemistry are strongly coupled in modulating the mechanical responses of the ECM.
4.2. Compression resistance of tumor ECM
Accumulated stresses arising from both volumetric growth and external environment may compress the blood vessels in
the ECM and impede the perfusion and delivery of anticancer drugs (Jain et al., 2014). Therefore, the compression resistance
of the tumor ECM plays a key role in cancer therapeutics. Besides, cancer cells generally have lower elastic moduli than
the corresponding normal cells (Fritsch et al., 2010; Katira et al., 2012). However, experiments showed that solid tumors
are often stiffer than the surrounding healthy tissues and can sustain higher compressive stresses (Jain et al., 2014). This
apparent contradiction is mainly attributed to the mechanical resistance capacity of tumor ECM, which possesses higher
elastic modulus due to its fibrous microstructure.
To examine the compression resistance of tumor ECMs, we first apply a confined compression model to determine the
mechanical and fluid transport parameters of a poroelastic tumor ECM (Netti et al., 20 0 0; Mow et al., 1984). The freely
swelling ECM is placed in a cylindrical chamber, as shown in Fig. 3(a). While its lateral expansion is confined by the cham-
ber, the ECM is assumed to be compressed in the thickness direction by a porous plate simulating the effects of surrounding
cancer cells or blood vessels, which can maintain chemical equilibrium with the external physiological salt solution and
allow interstitial fluid to flow freely between the ECM and the salt solution. In reality, tumor ECMs are not entirely confined
under in vivo conditions. Therefore, in parallel with the confined compression model, we also consider an unconfined ECM
disk compressed along the thickness direction but allowed to freely swell in the lateral direction, as shown in Fig. 3(b).
In the case of confined compression (Fig. 3a), the stretch ratios of the ECM in the thickness direction and the lateral
direction are
λ1 = λ0 (1 − εc ), λ2 = λ3 = λ0 , (65)
respectively, where ɛc is the nominal compressive strain and λ0 is the stretch ratio of the freely swelling ECM, which is
given by Eq. (62).
Fig. 4. Compressive stress–strain relations of ECM under confined and unconfined compressions.
The compressive stress applied on the porous platen is expressed as σc = σ1 + p, where σ 1 is the Cauchy stress in the
thickness direction. Using Eq. (27), we have

1 3 λ1
σc = ϕ1 E1 1 − − . (66)
3 I1 λ2 λ3
In combination with Eqs. (58) and (60), Eq. (66) becomes

1 3 λ1 kB T ϕ2 ϕ2 χ ϕ22
σ c = ϕ1 E1 1− + ln 1 − + +
3 λ2 λ3 νw
I1 J − ϕ1 J − ϕ1 (J − ϕ1 )2

z ϕ2 2
2
− kB T + 4c02 − 2c0 . (67)
ν2 J − ϕ 1 − ϕ 2
By inserting the stretch ratios in Eq. (65) into (67), we can determine the compressive stress–strain relation of the tumor
ECM under confined compression.
For the case of unconfined compression, we have
λ1 = λ0 (1 − εc ), λ2 = λ3 = λ, (68)
where λ can be determined by the equilibrium condition of Cauchy stress σ2 = −p in the lateral direction. Then the applied
compressive stress σc can be calculated by substituting the stretch ratios in Eq. (68) into (67).
For convenience, we use σc = |σ c | to represent the magnitude of the applied compressive stress. The stress–strain rela-
tion in Fig. 4 reveals its ability to sustain compression. However, the thin collagen fibrils and GAG chains in the ECM can bear
tensile stresses but can hardly endure compression. In fact, the GAGs attract a large amount of water molecules through os-
mosis, generating large GAG–water aggregations trapped by the collagen fibril network. It is the trapped water that endows
the tumor ECM with strong compression resistance. Note that in the case of unconfined compression, the tensile strength of
collagen fibrils also contributes to the compression resistance of the tumor ECM by restricting its radial extension. Therefore,
the integration of GAGs and the collagen network allows the tumor ECM to adapt to its complex microenvironments.
When the compressive strain ɛc is fixed, the corresponding compressive stress σ c under different ECM parameters can
be used to evaluate the compression resistance of the tumor ECM. Fig. 5(a) shows that the compression resistance is almost
proportional to the volume fraction ϕ 1 of collagen fibrils in both the confined and unconfined cases. As stated in Section
4.1, the collagen network limits the excessive expansion of the ECM and ensures the relatively high level of the osmotic
pressure. Fig. 5(b) illustrates that the compression resistance of ECM also depends on the amount of GAG chains, which
dictates the amount of water that can be trapped. However, the compression resistance does not rise monotonously with
increasing ϕ 2 . This is because an excess of GAG chains in the dry ECM may result in an excessively swollen matrix with a
low osmotic pressure.
Fig. 5(c) and (d) show that the compression resistance of tumor ECM varies with the volume fractions of collagen fibrils
and GAGs under confined and unconfined compression, respectively. Similar trends are observed under both compression
conditions. We find that the tumor ECM with either a low collagen or a low GAG level would have weak compression
resistance. A stiffer ECM can accumulate and transmit higher compressive stresses which may render vessel collapse. This
is consistent with the observation of Chauhan et al. (2013) that the blood vessels near the ECM with abundant collagen and
GAGs are greatly narrowed. Therefore, the compression resistance of an ECM results from a combined action of the GAG
chains and the collagen network.
Because of the significant influence of the negative charges in the GAG chains and free ions on the osmosis, they also
affect the compression resistance of tumor ECM. Fig. 5(e) shows that the charges carried by the GAG chains slightly enhance
Fig. 5. Compression resistance of ECM. Dependence of the compressive stress σ c on (a) the collagen volume fraction ϕ 1 , (b) the GAG volume fraction ϕ 2
in the case of confined and unconfined compressions. Phase diagrams of σ c under different ϕ 1 and ϕ 2 in the case of (c) confined and (d) unconfined
compression. Dependence of the compressive stress σ c on (e) the amount of negative charges z2 in each GAG chain, and (f) the normalized free ion
concentration c0∗ in the cases of confined and unconfined compression. Here, the compressive strain is taken as εc = 0.4.
the compression resistance due to the increase in osmotic pressure. When the concentration of free ions is relatively low,
its rise will increase the compression resistance in the ECM (Fig. 5(f)). However, further increase in the ion concentration
will weaken the osmotic pressure (as illustrated in Section 4.1), which, in turn, soften the ECM.
4.3. Drug delivery in tumor ECM
Tumor ECM has a significant influence on the delivery of cancer-fighting drugs, especially nanomedicines targeting
cancer cells. The movement of a drug particle in ECM depends on its size, shape, and charges, as well as the ECM’s physic-
ochemical properties (Albanese et al., 2012; Ernsting et al., 2013; Bertrand et al., 2014; Yu et al., 2016). Small therapeutic
nanomedicines with sizes on the order of a few nanometers can diffuse rapidly in tumor ECMs. However, the diffusion of
nanoparticles, such as liposomes and viruses whose size can be tens or even hundreds of nanometers, may be hindered by
a stiff and dense tumor ECM (Albanese et al., 2012). It has been recognized that the penetration of nanoparticles in tumors
depends on the volume fractions of the components in the ECM, especially the collagen and GAGs (Pluen et al., 2001; Jain
and Stylianopoulos, 2010).
Before reaching cancer cells, the drug nanoparticles in the blood circulation have to pass through the blood vessel walls
and penetrate the ECM. In desmoplastic tumors such as breast and pancreatic cancers, blood vessels are often surrounded
by a dense ECM (Smith et al., 2013). As observed in experiments (Yuan et al., 1994; Primeau et al., 2005; Perrault et al.,
2009), the ECM may pose a physiological barrier to nanomedicines. As shown in Fig. 6(a), liposomes with a radius of ∼45
nm can extravasate from blood vessels but cannot penetrate the tumor ECM and, consequently, they tend to concentrate in
Fig. 6. Drug delivery through blood vessels in a tumor. (a) Nanomedicines accumulated around blood vessels due to diffusional hindrance by ECM. Adapted
from Chauhan et al. (2011) with permission. (b) Model for a blood vessel embedded in a tumor ECM.
the perivascular region (Yuan et al., 1994). Motivated by these experiments, we examine the efficacy of drug delivery in an
ECM-surrounded blood vessel.
Consider a tumor ECM containing a cylindrical blood vessel, as shown in Fig. 6(b). The mechanical equilibrium of the
ECM in the cylindrical coordinate reads
dP1 1
+ (P1 − P2 ) = 0, (69)
dR R
where P1 and P2 are the first Piola-Kirchhoff stresses in the radial and circumferential directions, respectively.
For the axisymmetric problem, the transport Eqs. (39) and (40) in the reference configuration become
−2
1 ∂ ∂r ∂ μw D b ∂ μb
C˙ w = R k˜Cw Cw + Cb , (70)
R ∂R ∂R ∂R D0b ∂R
b
−2
1 ∂ ∂r Da ∂ μa D a ˜ ∂ μw D b ∂ μb
C˙ a = R Ca + 0 k Cw + Cb + ξa . (71)
R ∂R ∂R kB T ∂R Da ∂R D0b ∂R
b
Under the same boundary conditions defined in Section 4.1, we can determine the free-swelling state of the ECM model.
Since the swollen perivascular ECM is mechanically constrained by the blood vessel at the inner boundary and by the cell
aggregates at the outer boundary (Fig. 6), we fix the displacements of the inner and outer boundaries of the swollen ECM. In
the following, we will adjust the electrochemical boundary conditions including the fluid pressure, the drug concentration,
and the free ion concentration to analyze the drug delivery in the ECM.
Tumors often exhibit interstitial hypertension, as a result of the high permeability of tumor vessels and the lack of
functional lymphatic vessels in the interstitial space. Therefore, the interstitial fluid pressure (IFP) is elevated inside the
tumor and becomes almost equal to the microvascular pressure (MVP). The uniformly elevated IFP alleviates the pressure
gradients inside the tumor. Therefore, diffusion is the main mechanism of drug transport around the blood vessel while
the pressure gradient-dependent convection is weak (Jain and Stylianopoulos, 2010). Let p denote the fluid pressure
difference between MVP and IFP. The thickness of the ECM in the reference configuration HECM (= RECM − RV ) is 200μm
in this example (Fig. 6). Fig. 7 illustrates the drug diffusion in the ECM with p = 0. The concentration of therapeutic
agents will decay and even vanish in the blood circulation (Jain and Stylianopoulos, 2010). Hence, we assume that the drug
B
concentration in the blood circulation decays as cdrug 0
= cdrug exp(−kd t ), cdrug
0 being the initial drug concentration and kd the
∗
decay rate. Normalize the concentration of drugs in the ECM as cdrug 0
= cdrug /cdrug . The spatial distribution and evolution of
drug nanoparticles in the ECM are shown in Fig. 7(a), where we assume the drug concentration in the blood vessel to be
∗
time-independent, i.e., kd = 0. It shows that the drug concentration cdrug in the ECM reduces progressively with the distance
from the blood vessel. With increasing time, more and more drug nanoparticles will enter the ECM. In Fig. 7(b), we set
the decay rate of drug concentration in the blood circulation at kd = 0.001/day. At the initial stage, the nanomedicine is
transported into the ECM and the cellular space. After a period of time, however, the drug concentration in the perivascular
region will become lower than that in the cellular space, in agreement with the experiments (Kyle et al., 2007). Such a
reversed concentration gradient could cause the reflux of drugs, reducing therapeutic outcomes.
Experiments have shown that the physiochemical properties of therapeutic agents can influence their transport in ECM
(Chauhan and Jain, 2013). Here, we discuss the effect of drug size. It can be seen from Fig. 7(c) that the distribution of drugs
in the ECM is sensitive to their sizes. The amount of drug extravasating from the blood vessel decreases with increasing
drug size. In addition, large particles are difficult to penetrate the tumor ECM and thus heterogeneously concentrated
around the vessel (Yuan et al., 1994; Perrault et al., 2009). The heterogeneity and localization of drug distribution can
significantly limit the therapeutic efficacy. Since therapeutic agents can kill cancer cells only when they reach the cancer
cells, the drug concentration on the outer boundary of the ECM model, cdrug out , is here used as an index of drug efficacy
∗
Fig. 7. Distribution and evolution of the normalized concentration cdrug of a 10 nm drug particle in the ECM (a) without and (b) with drug decay in the
∗
blood vessel. (c) Drug concentration cdrug with different radii in the ECM. (d) Effect of nanoparticle radius rdrug on the normalized concentration c∗ on the
outer surface of the ECM model.
and a monitor of transport efficiency. The normalized drug concentration (cdrug out )∗ = cout /c0
drug drug
is written as c∗ for brevity
hereafter. In the following, therefore, we employ the normalized drug concentration c∗ to evaluate drug delivery in the
ECM. The effect of drug size on c∗ is shown in Fig. 7(d). Fig. 7(c) and (d) show that drug particles larger than a critical
size of about 30 nm cannot be efficiently transported through the ECM, which is consistent with experimental observations
(Jain and Stylianopoulos, 2010). For larger nanomedicines such as viruses and liposomes, the physical barrier due to the
ECM may become a primary restriction in oncology treatment.
Since the increased tumor IFP prevents the efficient uptake of therapeutic agents, it is of paramount importance to
find strategies to increase the transvascular transport in tumors. It has been demonstrated that in both animal models and
patients, this can be achieved by increasing systemic blood pressure (e.g., adding hypertensive molecules) or lowering IFP in
tumors (e.g., adding VEGF inhibitors) (Heldin et al., 2004). Both approaches are able to elevate the fluid pressure difference
p across the ECM. Fig. 8(a) illustrates the effects of the pressure difference on the drug delivery, suggesting that increasing
fluid pressure difference may result in a more efficient drug transport, in agreement with experimental observations (Netti
et al., 1995; Netti et al., 1999).
From the transport Eq. (40) or (71), we find that the transport of solute a in the ECM is driven by the concentration
gradient (∇ R ca )/ca , the overall fluid pressure gradient ∇ R , as well as the hydrodynamic hauling of other mobile particles.
In addition, the overall fluid pressure gradient and hydrodynamic interactions are also affected by solutes outside the ECM.
The fluid pressure difference between the inner and outer boundaries of the ECM model promotes the convection of the
solvent. Thereby, drug particles will move outwards under the driving action of ∇ R and the hydrodynamic forces of
solvent particles. Fig. 8(b) compares the drug transports under the following driving forces: (i) only the force arising from
the concentration gradient, (ii) the forces due to the concentration gradient and the hydrodynamics, and (iii) all forces due
to the concentration gradient, the hydrodynamic hauling, and the fluid pressure gradient. It can be seen from Fig. 8(b) that
p affects both the hydrodynamic hauling and the fluid pressure gradient ∇ R , and increasing p will greatly enhance
the driving effect of ∇ R and hence improve the drug transport in the ECM.
The blood vessels and cellular aggregates usually have different amounts of ions, proteins, and other colloids (Heldin et
al., 2004). For example, the abnormal physiological activities of cancer cells may release waste and modify the microen-
vironment, affecting the ion concentrations in the tumor tissue (Trédan et al., 2007). This diversity has been shown to
influence interstitial transport. Let c0∗ denote the difference between the normalized ion concentration in the blood vessels
and that in the cellular aggregates. We analyze the effects of c0∗ on drug transport in the ECM. As can be seen from Fig.
8(c), where we have set p = 0, increasing ion concentration in the blood slows down the transport of drug particles, while
Fig. 8. Evolution of c∗ with variations in (a) the pressure difference p and (c) the normalized free ion concentration difference c0∗ , where we have taken
rdrug = 10 nm. Influences of different driving forces on c∗ with the variations in (b) p and (d) c0∗ .
increasing ion concentration in the cellular space tends to accelerate the drug delivery in the ECM. Fig. 8(d) further reveals
the impact of driving forces on the drug transport. The hydrodynamic hauling serves as the main driving force in this case.
Eq. (53) gives the difference in electrochemical potential of the solvents as μw = νw ( p − 2kB T c0 ). Despite p = 0 in
this example, the ion concentration difference c0 leads to a distinct difference between the electrochemical potentials at
the two boundaries, resulting in a solvent flux. Since water tends to move along the direction of ion gradient, the higher
ion concentration in the blood will render a water flux toward the blood vessels. In this scenario, the hauling force of water
impedes the drug delivery.
4.4. Effects of ECM degradation
Experiments demonstrated that the co-injection of nanomedicines and ECM–degrading enzymes can enhance fluid
transport and reducing the mechanical resistance of ECM and, thus, improve the efficacy of chemotherapy (Jain and
Stylianopoulos, 2010; Jain, 2013). In this subsection, we dissect the effects of ECM degradation on its mechanical and
transport properties. For simplicity, we assume that the enzymatic degradation rates of collagen fibrils and GAG chains are
equal, i.e. ke10 = ke20 = ke0 . Hence, Eq. (51) reduces to
d˙1 = ke0 exp (−γ D1 )(1 − d1 ), (72)
d˙2 = ke0 (1 − d2 ), (73)

where γ is the mechanosensitive factor of collagen fibrils. Unless stated otherwise, the enzyme molecules are assumed
to be distributed uniformly in the ECM, yielding uniform degradations of collagen fibrils and GAG chains. According to
experiments (Mckee et al., 2006; Mok et al., 2007; Stylianopoulos et al., 2012; Provenzano et al., 2012), we account for
three ECM normalization strategies, including collagen degradation, GAG degradation, and their combinations.
4.4.1. Stress alleviation

The effects of ECM degradation on the compression resistance are first considered, as shown in Fig. 9. For the three ECM
normalization strategies, Fig. 9(a) and (b) illustrate the results for the confined and unconfined compressions of the ECM
(Fig. 3), respectively, where the external compressive strain is set as 0.4. In the calculations, we assume that the fluid con-
stituents in the system are at the electrochemical equilibrium state throughout the degradation process. As aforementioned,
Fig. 9. Evolution of the steady–stated σ c under different ECM degradation treatments in the cases of (a) confined and (b) unconfined compression with
compressive strain εc = 0.4 and (ke0 )∗ = 1. (c) Dependence of the steady–stated σ c on the normalized enzymatic degradation constant (ke0 )∗ and the
mechanosensitive factor γ under collagen degradation.
the osmotic pressure is balanced by the elasticity of the collagen network and the externally applied compressive load.
Under a given compressive strain, the stresses in the network will drop with the enzymatic dissolution of collagen fibrils.
Correspondingly, in the case of confined compression, the externally applied compressive stress σ c will rise to counteract the
osmotic pressure (Fig. 9(a)). In the case of unconfined compression, however, the degradation of the collagen network would
lead to a decrease in σ c (Fig. 9(b)), since the collagen degradation softens the ECM and weakens its compression resistance.
In addition, it can be seen from Fig. 9(a) and (b) that GAG degradation can reduce the compression resistance of tumor
ECM in both confined and unconfined compression. Since GAG chains are responsible for the ECM osmosis, the decrease
in GAG content is companied with a drop of the osmotic pressure, corresponding to a decrease in the externally applied
compressive stress. In the case of confined compression, the applied stress σ c might even change from compressive to
tensile with the degradation of GAG chains, as shown by the red curve in Fig. 9(a). The degradation of GAG chains reduces
the osmotic pressure in the ECM and thus decreases the ECM volume. When a sufficient fraction of GAG chains have
been dissolved, the stress-free volume of the ECM could become even smaller than the confined chamber and, thus, an
externally applied tensile stress is needed to maintain the constant volume of the confined ECM. For the case of unconfined
compression, the axial stress also decreases with the degradation of GAG chains. However, due to the free motion of
lateral boundaries in this case, the reduction of σ c is significantly slower than that in the confined compression. The stress
alleviations with the degradation of collagen and GAGs are consistent with experiments (Stylianopoulos et al., 2012).
To address the influence of enzyme reaction efficiency on the ECM degradation, we define the normalized enzymatic
degradation rate (ke0 )∗ = ke0 /(ke0 )ref , where (ke0 )ref = 6 × 10−4 s−1 . A larger enzyme degradation constant ke0 will lead to a
higher degradation degree and undermine the compression resistance of the interstitial matrix more efficiently, as shown
in Fig. 9(c). Another factor that affects the enzymatic reaction is the mechanosensitivity factor of collagen degradation,
γ . Under the same enzyme concentration and mechanical load, a larger γ will render a slower collagen dissolution since
the elevated γ tends to prevent the ECM from enzymatic degradation, as verified in Fig. 9(c). If γ is sufficiently large, the
stretched collagen fibril can be hardly dissolved by enzyme molecules, regardless of the enzyme reaction efficiency.
In the above analysis, we have assumed electrochemical equilibrium of fluids, corresponding to the steady-state ECM
normalization. This assumption holds for most practical situations in which the ECM degradation is relative slow. In the
case of a nonequilibrium electrochemical state, the degradation process may involve the transports of both mobile fluid and
Fig. 10. (a) Evolution of transient state stress σ c in different ECM normalization treatments in the case of confined compression with εc = 0.4 and (ke0 )∗ =
1. Distributions and evolutions of (b) the normalized solvent flux Qw ∗
and (c) the overall fluid pressure with collagen degradation. (d) Evolution of on
the inner and outer boundaries with collagen degradation. (e) Evolution of σ c under different (ke0 )∗ in the case of collagen degradation. (f) Effect of (ke0 )∗
on σ c with a uniform collagen degradation degree.
solid particles (e.g., water) in the matrix. In this case, the ECM thickness HECM may affect the mechanical response of the
ECM. Fig. 10 shows the transient alleviation of stresses in the ECM under the condition of confined compression defined in
Fig. 3(a), where we have set HECM = 300 μm. Its comparison with Fig. 9(a) shows that the transient stress distribution in
the ECM normalization is quite different from that at the steady state, especially in the treatment of collagen degradation.
With the degradation of collagen fibrils, the ECM subject to a given compressive strain bears a distinctly larger steady–state
compressive stress (Fig. 9(a)). However, during the catalytic reaction of collagens, the transient blocking stress will exhibit a
non-monotonic change: it first drops rapidly and then rises gradually, as shown in Fig. 10(a). This non-monotonic variation
in stress may be attributed to the effect of fluid movements with collagen dissolution, which leads to a decrease in the
overall volume of the ECM. Under the incompressibility condition, a reduction in the ECM volume induced by collagen
degradation is compensated by infusion of fluid constituents, as verified in Fig. 10(b). In the present numerical example,
we have set the bottom boundary of the ECM model to be flux-free and, thus, the solvent particles can flow into the ECM
only through the top boundary (Fig. 3(a)). At the early stage of degradation, water infuses into the matrix, generating a
flux on the ECM surface. Limited by the fluid transport efficiency, the fluid constituents are heterogeneously distributed
and highly located in the vicinity of the free boundaries, rendering non-uniform stress distribution in the beginning stage
Fig. 11. Evolution of normalized drug concentration c∗ under different ECM normalization treatments with (a) ϕ1 = 0.1 and ϕ2 = 0.01, and (b) ϕ1 = ϕ2 =
0.02. (c) Effects of ECM normalization on c∗ with the variation in p. (d) Distribution and evolution of overall fluid pressure in the ECM under GAG
degradation.
of degradation. The distributions of fluid constituents and stress will become more and more uniform as time elapses, as
shown in Fig. 10(b). The decrease in the ECM volume induced by collagen dissolution leads to the decline in the overall
fluid pressure . Corresponding to the non-monotonic stress variation, also drops steeply at the initial stage and rises
again due to water influx, as shown in Fig. 10(b–d). The blocking stress σ c and the overall fluid pressure share the same
trend of variation, and the difference in their magnitudes stems from the contribution of the collagen network.
Fig. 10(e) and (f) further illustrate the effect of enzymatic degradation rate ke0 on the blocking stress σ c . With increasing
degradation rate, both the blocking stress and the overall fluid pressure undergo a sharp decline in the early stage of
degradation, indicating that the stresses are sensitive to the transient volume change induced by ECM degradation. The
rate of volume change originates from a competition between polymer degradation and fluid perfusion: a higher degra-
dation rate of collagen fibrils causes a faster decrease in the ECM volume, while the fluid perfusion rate remains almost
unchanged. As a result, the transient stress will drop steeply. If the effect of mechanical regulation is neglected, the collagen
degradation factor has an explicit expression, d1 = 1 − exp(−ke0 td ). In Fig. 10(f), we assume that ke0 td is a constant and thus
the volume loss of collagen is unchanged. Fig. 10(f) clearly shows that a larger ke0 can lead to a decrease in the transient
compressive stress. For enzymatic degradation, the loss rate of collagen volume decreases exponentially with time, that is,
d˙1 = ke0 exp(−ke0 td ), and the ECM may be gradually filled by the continuously perfused water.
4.4.2. Drug transport in degraded ECM

Experiments found that therapeutic nanoparticles with radius larger than 30 nm can hardly penetrate a dense tumor
ECM, but their delivery efficiency can be substantially improved by ECM normalization treatment (Jain and Stylianopoulos,
2010; Jain, 2013). We now consider how ECM normalization facilitates drug transport. The effects of ECM normalization on
drug transport are shown in Fig. 11. Obviously, the degradation treatment yields a better efficacy of nanomedicine in the de-
graded, sparse ECM than that in the original dense case, as shown in Fig. 11(a) and (b), suggesting that the ECM degradation
of desmoplastic tumors is an efficient strategy to enhance the transport of cancer-fighting drug. After the ECM degradation,
larger drug nanoparticles can go beyond the perivascular region and get in contact with the targeted cancer cells. Though
the degradation of either collagen or GAGs can enhance drug delivery, a combined strategy may achieve an optimal outcome.
We further examine the effects of ECM normalization on drug transport under different external environments. Fig. 11(c)
identifies the effect of ECM normalizations on the convective transport of drugs driven by the fluid pressure gradient. It
is seen that the degradation of GAG chains renders an apparent enhancement in the drug delivery. It is known from Eqs.
(28), (56), and (57) that = μw /νw +, indicating that the overall fluid pressure gradient ∇ R contains the contributions
Fig. 12. Effects of limited enzyme transport in different ECM normalization treatments. (a) Dependence of normalized drug concentration c∗ on the enzyme
radius re , where (ke0 )∗ = 0.1. (b) Distributions of overall fluid pressure , where re = 10nm. (c) Effect of the pressure difference p on c∗ , where re = 10nm.
(d) Phase diagrams of c∗ with respect to (ke0 )∗ and re under GAG degradation.
from the osmotic pressure gradient ∇ R and the electrochemical potential gradient of solvent ∇ R μw . When MVP is higher
than IFP, there exists a chemical potential gradient of solvents in the ECM. In this case, the perivascular region of the ECM
will trap more water and swell more. Since osmosis tends to eliminate the inhomogeneity of the electrochemical potential
in the solvents, the osmotic pressure will also become lower in the perivascular region. Thus, the direction of ∇ R is
opposite to ∇ R μw . However, with the dissolution of GAGs, the osmotic pressure in the ECM drops, ∇ R levels off, and the
fluid pressure gradient ∇ R elevates correspondingly. Fig. 11(d) shows the evolution of the overall fluid pressure in the
ECM undergoing GAG degradation, demonstrating that GAG degradation can elevate ∇ R . In addition, GAG degradation can
improve the permeability and enhance the fluid convection. The combination of these effects can yield a higher efficacy of
nanomedicine delivery. Compared with GAG degradation, collagen normalization only results in a modest improvement of
drug delivery. The reason is that the decreased collagen content only enhances the diffusivity of the ECM, which is in fact
less important in drug transport when p is high.
Finally, we discuss the transport efficiency of enzyme molecules on drug delivery. In the above analysis, we have as-
sumed that the enzyme concentrations are constant during the degradation process. In fact, the enzyme solutions injected
into the ECM also undergo a tempo-spatial evolution. In many situations, the sizes of enzymes are not small enough to
freely penetrate through the dense matrix, and thus their concentrations are also nonhomogeneous, eliciting uneven ECM
degradation. Fig. 12(a) illustrates the influence of enzyme size on drug delivery. It is found that larger enzyme molecules are
less helpful for improving drug transport because of their uneven distribution. Besides, it is interesting to find that the ECM
normalization treatment may reduce drug transport if the enzyme size is very large, as shown in Fig. 12(a). It is easy to
speculate that the main impediment to the drug delivery is the osmotic pressure gradient induced by the inhomogeneous
ECM degradation. Actually, Fig. 12(b) indicates that the ECM degradation treatments will generate an osmotic pressure
gradient in the opposite direction of drug transport, making the drug penetration slower. Taken together, it is concluded that
the enzyme delivery efficiency is important for drug delivery, and the enzyme size should be optimized in a certain range
to ensure the efficacy of ECM normalization. Smaller enzyme molecules are preferential due to their efficient transport and
homogeneous distributions in the matrix. Raising the level of MVP seems to be a strategy to overcome the unfavorable
effects of enzyme delivery. As shown in Fig. 12(c), provided that the fluid pressure difference is larger than a threshold
(e.g., 1 kPa), the normalization treatment will benefit the drug transport. Besides, since enzymatic degradation may have
adverse effects on the drug transport under limited enzyme delivery, an increase in the enzymatic degradation rate will
decrease the drug delivery efficacy, as verified in Fig. 12(d). Smaller enzyme molecules will have a relatively homogenous
distribution in the ECM, enhancing the efficacy of drug delivery.
5. Conclusions
The mechanical and chemical properties of tumor ECM have great impacts on the microenvironments of tumor and
the efficacy of anticancer therapeutics. In this work, we have established a chemomechanical thermodynamic theory to
investigate a number of fundamental and physiologically relevant properties of tumor ECM, such as compression resistance,
fluid transport, and chemical degradation. Based on the microstructural and chemomechanical considerations of the network
of collagen fibrils and GAG chains, we have developed a constitutive law of tumor ECM treated as a polyelectrolyte. A
generalized transport equation coupling fluid diffusion, convection, and reaction is derived to describe the movements of
fluid constituents in the ECM. In addition, we have proposed a mechanosensitive enzymatic degradation law to rationalize
the ECM normalization treatment. This theory have been validated by comparing with relevant experiments. The main
conclusions drawn from this work are as follows.
(1) The compression resistance of a tumor ECM originates from its osmosis, contributed by both the collagen network
and GAG chains. GAG chains imbibe water, while the collagen network limits the excessive ECM swelling and ensures
a higher osmotic pressure to resist the applied compressive stress.
(2) The uniform IFP limits fluid diffusion, while rising MVP improves fluid convection and thus facilitates drug delivery
in the ECM. Besides, the difference of free ion concentrations in blood and interstitial fluids can also affect the mass
transport in the ECM.
(3) Degradation treatments can effectively soften the ECM and alleviate the compressive stresses in tumors, so as to relax
the blood vessels and restore blood flows in the tumor. Furthermore, ECM normalization can improve the drug deliv-
ery through both diffusion and convection. However, the limited transport of enzymes with large sizes may impede
the transport of nanomedicine.
Finally, it is pointed out that the present study aims at providing a generic theoretical model for solid tumor ECMs.
However, to comprehensively characterize the tumor microenvironment and assess the outcome of a prescribed treatment,
one needs a full description of both the extracellular matrix and cells in the tumor. In addition, we have only addressed
the size effects of nanomedicine on its delivery by assuming that the particles are close to spheres, without considering
the influences of shape and surface chemistry of drugs (Chauhan and Jain, 2013; Yu et al., 2016). These issues merit further
detailed investigation to develop efficient therapeutic strategies against cancer.
Acknowledgments
Supports from the National Natural Science Foundation of China (Grant Nos. 11432008, 11620101001, and 11672161),
Tsinghua University (20121087991 and 20151080441), and the Thousand Young Talents Program of China are acknowledged.
Appendix. Permeability of matrix
The Carmen–Kozeny (CK) equation predicts the hydraulic drag due to viscous interaction between the matrix and the
flowing fluid (Levick, 1987; Swartz and Fleury, 2007). According to this equation, the hydraulic permeability of component
m in a tumor ECM can be written as
1 δm 3
km = , (A1)
ηvis Gm Sm2
where δ m is the porosity of the meshwork comprising only the component m, Sm the wetted surface area of fibrils per unit
volume, and Gm the Kozeny factor. For convenience, the porosity δ m is re-expressed as δm = 1 − ϕˆ m . Considering the orienta-
//
tion of fibrils, Gm can be estimated from the contribution of fibrils perpendicular (G⊥
m ) and parallel (Gm ) to the flow, that is,
2 ⊥ 1 //
Gm = G + G , (A2)
3 m 3 m
where
2δm3
G⊥
m = # $, (A3)
1 − ( 1 − δm )
2
(1 − δm ) − ln (1 − δm ) − 1+(1−δm )
2
2δm
3
G//
m = . (A4)
(1 − δm )[−2 ln(1 − δm ) − 3 + 4(1 − δm ) − (1 − δm )2 ]
Besides, for fibrils with circular cross-section, we have ϕˆ m /Sm = rm /2. Then Eq. (A1) becomes
−1 −1
1 3rm 2
−1 1 − ϕˆ m
2
km = − [2 ln ϕˆ m + (ϕˆ m − 1 )(ϕˆ m − 3 )] + 2 ln ϕˆ m + . (A5)
ηvis 8ϕˆ m 1+ϕˆ m2
References
Abu Al-Rub, R.K., Darabi, M.K., 2012. A thermodynamic framework for constitutive modeling of time- and rate-dependent materials. Part I: theory. Int. J.
Plast. 34, 61–92.
Albanese, A., Tang, P.S., Chan, W.C.W., 2012. The effect of nanoparticle size, shape, and surface chemistry on biological systems. Annu. Rev. Biomed. Eng. 14,
1–16.
Amsden, B., 1998. Solute diffusion within hydrogels. Mechanisms and models. Macromolecules 31, 8382–8395.
Arruda, E.M., Boyce, M.C., 1993. A three-dimensional constitutive model for the large stretch behavior of rubber elastic materials. J. Mech. Phys. Solids 41,
389–412.
Ateshian, G.A., Likhitpanichkul, M., Hung, C.T., 2006. A mixture theory analysis for passive transport in osmotic loading of cells. J. Biomech. 39, 464–475.
Aukland, K., Nicolaysen, G., 1981. Interstitial fluid volume: local regulatory mechanisms. Physiol. Rev. 61, 556–643.
Bertrand, N., Wu, J., Xu, X., Kamaly, N., Farokhzad, O.C., 2014. Cancer nanotechnology: the impact of passive and active targeting in the era of modern cancer
biology. Adv. Drug Delivery Rev. 66, 2–25.
Biot, M.A., 1941. General theory of three-dimensional consolidation. J. Appl. Phys. 12, 155–164.
Biot, M.A., 1972. Theory of finite deformations of porous solids. Indiana Univ. Math. J. 21, 597–620.
Brown, E., McKee, T., diTomaso, E., Pluen, A., Seed, B., Boucher, Y., Jain, R.K., 2003. Dynamic imaging of collagen and its modulation in tumors in vivo using
second–harmonic generation. Nat. Med. 9, 796–800.
Camp, R.J., Liles, M., Beale, J., Saeidi, N., Flynn, B.P., et al., 2011. Molecular mechanochemistry: low force switch slows enzymatic cleavage of human type I
collagen monomer. J. Am. Chem. Soc. 133, 4073–4078.
Chang, S.W., Flynn, B.P., Ruberti, J.W., Buehler, M.J., 2012. Molecular mechanism of force induced stabilization of collagen against enzymatic breakdown.
Biomaterials 33, 3852–3859.
Chauhan, V.P., Jain, R.K., 2013. Strategies for advancing cancer nanomedicine. Nat. Mater. 12, 958–962.
Chauhan, V.P., Martin, J.D., Liu, H., Lacorre, D.A., Jain, S.R., et al., 2013. Angiotensin inhibition enhances drug delivery and potentiates chemotherapy by
decompressing tumour blood vessels. Nat. Commun. 4, 2516.
Chauhan, V.P., Stylianopoulos, T., Boucher, Y., Jain, R.K., 2011. Delivery of molecular and nanoscale medicine to tumors: transport barriers and strategies.
Annu. Rev. Chem. Biomol. 2, 281–298.
Cheng, X., Petsche, S.J., Pinsky, P.M., 2015. A structural model for the in vivo human cornea including collagen-swelling interaction. J. R. Soc. Interface 12,
20150241.
Coussy, O., 2004. Poromechanics. John Wiley & Sons, Ltd.
Cowin, S.C., Cardoso, L., 2012. Mixture theory-based poroelasticity as a model of interstitial tissue growth. Mech. Mater. 44, 47–57.
Cukier, R.I., 1984. Diffusion of Brownian spheres in semidilute polymer solutions. Macromolecules 17, 252–255.
DuFort, C.C., DelGiorno, K.E., Carlson, M.A., Osgood, R.J., Zhao, C., et al., 2016. Interstitial pressure in pancreatic ductal adenocarcinoma is dominated by a
gel-fluid phase. Biophys. J. 110, 2106–2119.
Ernsting, M.J., Murakami, M., Roy, A., Li, S.D., 2013. Factors controlling the pharmacokinetics, biodistribution and intratumoral penetration of nanoparticles.
J. Controlled Release 172, 782–794.
Flory, P.I., 1942. Thermodynamics of high polymer solutions. J. Chem. Phys. 10, 51–61.
Flynn, B.P., Tilburey, G.E., Ruberti, J.W., 2013. Highly sensitive single-fibril erosion assay demonstrates mechanochemical switch in native collagen fibrils.
Biomech. Model. Mechanobiol. 12, 291–300.
Frantz, C., Stewart, K.M., Weaver, V.M., 2010. The extracellular matrix at a glance. J. Cell Sci. 123, 4195–4200.
Fritsch, A., Höckel, M., Kiessling, T., Nnetu, K.D., Wetzel, F., Zink, M., Käs, J.A., 2010. Are biomechanical changes necessary for tumor progression? Nat. Phys.
6, 730–732.
Gibbs, J.W., 1878. The Scientific Papers of J. Willard Gibbs, p. 184 201, 215.
Gilkes, D.M., Semenza, G.L., Wirtz, D., 2014. Hypoxia and the extracellular matrix: drivers of tumor metastasis. Nat. Rev. Cancer 14, 430–439.
Gu, W.Y., Lai, W.M., Mow, V.C., 1998. A mixture theory for charged-hydrated soft tissues containing multi-electrolytes: passive transport and swelling
behaviors. J. Biomech. Eng. 120, 169–180.
Huggins, M.L., 1942. Some properties of solutions of long-chain compounds. J. Phys. Chem. 46, 151–158.
Heldin, C.H., Rubin, K., Pietras, K., Östman, A., 2004. High interstitial fluid pressure— an obstacle in cancer therapy. Nat. Rev. Cancer 4, 806–813.
Hong, W., Zhao, X., Suo, Z., 2010. Large deformation and electrochemistry of polyelectrolyte gels. J. Mech. Phys. Solids 58, 558–577.
Hong, W., Zhao, X., Zhou, J., Suo, Z., 2008. A theory of coupled diffusion and large deformation in polymeric gels. J. Mech. Phys. Solids 56, 1779–1793.
Jain, R.K., 1987. Transport of molecules in the tumor interstitium: a review. Cancer Res. 47, 3039–3051.
Jain, R.K., 1990. Physiological barriers to delivery of monoclonal antibodies and other macromolecules in tumors. Cancer Res. 50, 814–819.
Jain, R.K., 2013. Normalizing tumor microenvironment to treat cancer: bench to bedside to biomarkers. J. Clin. Oncol. 31, 2205–2218.
Jain, R.K., Martin, J.D., Stylianopoulos, T., 2014. The role of mechanical forces in tumor growth and therapy. Annu. Rev. Biomed. Eng. 16, 321–346.
Jain, R.K., Stylianopoulos, T., 2010. Delivering nanomedicine to solid tumors. Nat. Rev. Clin. Oncol. 7, 653–664.
Katira, P., Zaman, M.H., Bonnecaze, R.T., 2012. How changes in cell mechanical properties induce cancerous behavior. Phys. Rev. Lett. 108, 028103.
Kuhl, E., Garikipati, K., Arruda, E.M., Grosh, K., 2005. Remodeling of biological tissue: mechanically induced reorientation of a transversely isotropic chain
network. J. Mech. Phys. Solids 53, 1552–1573.
Kyle, A.H., Huxham, L.A., Yeoman, D.M., Minchinton, A.I., 2007. Limited tissue penetration of taxanes: a mechanism for resistance in solid tumors. Clin.
Cancer Res. 13, 2804–2810.
Lai, W.M., Hou, J.S., Mow, V.C., 1991. A triphasic theory for the swelling and deformation behaviors of articular-cartilage. J. Biomech. Eng. 113, 45–258.
Levick, J.R., 1987. Flow through interstitium and other fibrous matrices. Q. J. Exp. Physiol. 72, 409–438.
Liu, J., Liao, S., Diop-Frimpong, B., Chen, W., Goel, S., Naxerova, K., et al., 2012. TGF-β blockade improves the distribution and efficacy of therapeutics in
breast carcinoma by normalizing the tumor stroma. Proc. Natl. Acad. Sci. U.S.A. 109, 16618–16623.
Lu, P., Weaver, V.M., Werb, Z., 2012. The extracellular matrix: a dynamic niche in cancer progression. J. Cell Biol. 196, 395–406.
Magzoub, M., Jin, S., Verkman, A.S., 2008. Enhanced macromolecule diffusion deep in tumors after enzymatic digestion of extracellular matrix collagen and
its associated proteoglycan decorin. FASEB J. 22, 276–284.
Marcombe, R., Cai, S., Hong, W., Zhao, X., Lapusta, Y., Suo, Z., 2010. A theory of constrained swelling of a pH-sensitive hydrogel. Soft Matter 6, 784–793.
Masaro, L., Zhu, X.X., 1999. Physical models of diffusion for polymer solutions, gels and solids. Prog. Polym. Sci. 24, 731–775.
McKee, T.D., Grandi, P., Mok, W., Alexandrakis, G., Insin, N., 2006. Degradation of fibrillar collagen in a human melanoma xenograft improves the efficacy of
an oncolytic herpes simplex virus vector. Cancer Res. 66, 2509–2513.
Metters, A., Hubbell, J., 2005. Network formation and degradation behavior of hydrogels formed by Michael-type addition reactions. Biomacromolecules 6,
290–301.
Miao, L., Lin, C.M., Huang, L., 2015. Stromal barriers and strategies for the delivery of nanomedicine to desmoplastic tumors. J. Controlled Release 219,
192–204.
Mok, W., Boucher, Y., Jain, R.K., 2007. Matrix metalloproteinases-1 and -8 improve the distribution and efficacy of an oncolytic virus. Cancer Res. 67,
10664–10668.
Mow, V.C., Holmes, M.H., Lai, W.M., 1984. Fluid transport and mechanical properties of articular cartilage: a review. J. Biomech. 17, 377–394.
Netti, P.A., Baxter, L.T., Boucher, Y., Skalak, R., Jain, R.K., 1995. Time-dependent Behavior of interstitial fluid pressure in solid tumors: implications for drug
delivery. Cancer Res. 55, 5451–5458.
Netti, P.A., Berk, D.A., Swartz, M.A., Grodzinsky, A.J., Jain, R.K., 20 0 0. Role of extracellular matrix assembly in interstitial transport in solid tumors. Cancer
Res. 60, 2497–2503.
Netti, P.A., Hamberg, L.M., Babich, J.W., Kierstead, D., Graham, W., et al., 1999. Enhancement of fluid filtration across tumor vessels: implication for delivery
of macromolecules. Proc. Natl. Acad. Sci. U.S.A. 96, 3137–3142.
Palmer, J.S., Boyce, M.C., 2008. Constitutive modeling of the stress–strain behavior of F-actin filament networks. Acta Biomater. 4, 597–612.
Pan, Y., Zhong, Z., 2014. A nonlinear constitutive model of unidirectional natural fiber reinforced composites considering moisture absorption. J. Mech. Phys.
Solids 69, 132–142.
Perentes, J.Y., McKee, T.D., Ley, C.D., Mathiew, H., Dawson, M., et al., 2009. In vivo imaging of extracellular matrix remodeling by tumor associated fibroblasts.
Nat. Methods 6, 143–145.
Perrault, S.D., Walkey, C., Jennings, T., Fischer, H.C., Chan, W.C.W., 2009. Mediating tumor targeting efficiency of nanoparticles through design. Nano Lett. 9,
1909–1915.
Perumal, S., Antipova, O., Orgel, J.P.R.O., 2008. Collagen fibril architecture, domain organization, and triple-helical conformation govern its proteolysis. Proc.
Natl. Acad. Sci. U.S.A. 105, 2824–2829.
Pluen, A., Boucher, Y., Ramanujan, S., McKee, T.D., Gohongi, T., et al., 2001. Role of tumor–host interactions in interstitial diffusion of macromolecules: cranial
vs. subcutaneous tumors. Proc. Natl. Acad. Sci. U.S.A. 98, 4628–4633.
Primeau, A.J., Rendon, A., Hedley, D., Lilge, L., Tannock, I.F., 2005. The distribution of the anticancer drug doxorubicin in relation to blood vessels in solid
tumors. Clin. Cancer Res. 11, 8782–8788.
Provenzano, P.P., Cuevas, C., Chang, A.E., Goel, V.K., Von Hoff, D.D., Hingorani, S.R., 2012. Enzymatic targeting of the stroma ablates physical barriers to
treatment of pancreatic ductal adenocarcinoma. Cancer Cell 21, 418–429.
Rajagopal, K.R., Srinivasa, A.R., 2004. On thermomechanical restrictions of continua. Proc. R. Soc. A 460, 631–651.
Rajagopal, K.R., Srinivasa, A.R., Wineman, A.S., 2007. On the shear and bending of a degrading polymer beam. Int. J. Plast. 23, 1618–1636.
Ramanujan, S., Pluen, A., McKee, T.D., Brown, E.B., Boucher, Y., Jain, R.K., 2002. Diffusion and convection in collagen gels: implications for transport in the
tumor interstitium. Biophys. J. 83, 1650–1660.
Ruberti, J.W., Hallab, N.J., 2005. Strain-controlled enzymatic cleavage of collagen in loaded matrix. Biochem. Biophys. Res. Commun. 336.
Smith, N.R., Baker, D., Farren, M., Pommier, A., Swann, R., Wang, X., et al., 2013. Tumor stromal architecture can define the intrinsic tumor response to
VEGF-targeted therapy. Clin. Cancer Res. 19, 6943–6956.
Soulhat, J., Buschmann, M.D., Shirazi-Adl, A., 1999. A fibril-network-reinforced biphasic model of cartilage in unconfined compression. J. Biomech. Eng. 121,
340–347.
Stylianopoulos, T., Barocas, V.H., 2007. Volume-averaging theory for the study of the mechanics of collagen networks. Comput. Methods Appl. Mech. Engrg.
196, 2981–2990.
Stylianopoulos, T., Martina, J.D., Chauhan, V.P., Jain, S.R., Diop-Frimpong, B., et al., 2012. Causes, consequences, and remedies for growth-induced solid stress
in murine and human tumors. Proc. Natl. Acad. Sci. U.S.A. 109, 15101–15108.
Sugahara, K.N., Teesalu, T., Karmali, P.P., Kotamraju, V.R., Agemy, L., 2010. Coadministration of a tumor-penetrating peptide enhances the efficacy of cancer
drugs. Science 328, 1031–1035.
Swartz, M.A., Fleury, M.E., 2007. Interstitial flow and its effects in soft tissues. Annu. Rev. Biomed. Eng. 9, 229–256.
Trédan, O., Galmarini, C.M., Patel, K., Tannock, I.F., 2007. Drug resistance and the solid tumor microenvironment. J. Natl. Cancer Inst. 99, 1441–1454.
Xue, S.L., Li, B., Feng, X.Q., Gao, H., 2016. Biochemomechanical poroelastic theory of avascular tumor growth. J. Mech. Phys. Solids 94, 409–432.
Yu, M.R., Wang, J.L., Yang, Y.W., Zhu, C.L., Su, Q., Guo, S.Y., Sun, J.S., Gan, Y., Shi, X.H., Gao, H., 2016. Rotation-facilitated rapid transport of nanorods in
mucosal tissues. Nano Lett. 16, 7176–7182.
Yuan, F., Leunig, M., Huang, S.K., Berk, D.A., Papahadjopoulos, D., Jain, R.K., 1994. Microvascular permeability and interstitial penetration of sterically stabi-
lized (stealth) liposomes in a human tumor xenograft. Cancer Res. 54, 3352–3356.
Ziegler, H., 1977. An Introduction to Thermodynamics. North-Holland Pub., Amsterdam.

A Non-Equilibrium Thermodynamic Model For Tumor Extracellular Matrix With Enzymatic Degradation

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Non-Equilibrium Thermodynamic Model For Tumor Extracellular Matrix With Enzymatic Degradation

Uploaded by

Copyright:

Available Formats

Journal of the Mechanics and Physics of Solids 104 (2017) 32–56

Contents lists available at ScienceDirect

Journal of the Mechanics and Physics of Solids

A non-equilibrium thermodynamic model for tumor

Force and deformation

Fig. 1. Schematic of ECM in a solid tumor.

2. Theoretical framework of poroelastic solids

Its differential form is

3. Tumor ECM model

3.1. Constitutive equations

By distinguishing between solvent and solutes, Eq. (33) can be rewritten as

where C = FT · F and k˜ is deﬁned as

3.3. Degradation law

4. Examples and discussions

Symbol Description Value Source

4.1. Free swelling of tumor ECM

4.2. Compression resistance of tumor ECM

4.3. Drug delivery in tumor ECM

4.4. Effects of ECM degradation

d˙1 = ke0 exp (−γ D1 )(1 − d1 ), (72)

d˙2 = ke0 (1 − d2 ), (73)

4.4.1. Stress alleviation

4.4.2. Drug transport in degraded ECM

Appendix. Permeability of matrix

You might also like