394
4, DECEMBER 2012
Control of Low-Density Lipoprotein
Concentration in the Arterial Wall by Proportional
Drug-Encapsulated Nanoparticles
Alireza Rowhanimanesh and Mohammad-R. Akbarzadeh-T. , Senior Member, IEEE
Abstract—Atherosclerosis, or hardening of the arteries, is one
of the major causes of death in humans. High accumulation
of Low-Density Lipoprotein (LDL) macromolecules within the
arterial wall plays a critical role in initiation and development
of atherosclerotic plaques. This paper proposes a proportional
drug-encapsulated nanoparticle (PDENP) that utilizes a simple
piecewise-proportional controller to realize swarm feedback
control of LDL concentration in the interior of the arterial wall.
In contrast to the competing strategies on nanorobotics, PDENPs
carry simpler hardware architecture in order to be more reason-
ably realized technologically as well as to penetrate the interior
arterial wall. Furthermore, in contrast to the existing targeted
DENPs that usually target the surface proteins of atherosclerotic
plaque, the proposed PDENPs directly sense the LDL level in
the arterial walls. Hence, they can diagnose abnormal LDL
accumulation before plaque formation, prevent critical growth
of atherosclerotic plaques, while considerably reducing the un-
wanted drug side effects in healthy tissue. Simulation results on a
well-known mathematical model of the arterial wall demonstrate
that the proposed approach successfully reduces the LDL level to
a desired value in the arterial wall of a patient with very high LDL
level. Also, the mass of the released drug by PDENPs in a healthy
wall is 11 times less than its corresponding value in an unhealthy
wall.
Index Terms—Atherosclerosis, LDL concentration, nanoparti-
cles, nonlinear control, proportional control, swarm control.
I. INTRODUCTION
A THEROSCLEROSIS, or hardening of the arteries, is one
of the commonest causes of death in humans. Medical
investigations have shown that an abnormally high accumula-
tion of LDL (low-density lipoprotein) macromolecules within
the arterial wall plays a critical role in initiation and develop-
ment of atherosclerotic plaques [1]. In recent years, remarkable
advances in nanomedicine have made it possible to manufacture
drug-encapsulated nanoparticles (DENPs) which are capable of
Manuscript received February 01, 2012; revised August 03, 2012; accepted
August 26, 2012. Date of publication September 19, 2012; date of current ver-
sion November 29, 2012. Asterisk indicates corresponding author.
A. Rowhanimanesh is with the Center of Excellence on Soft Computing and
Intelligent Information Processing (SCIIP) and Department of Electrical (Con-
trol) Engineering, Ferdowsi University of Mashhad, Mashhad, Iran (e-mail:
rowhanimanesh@ieee.org).
*M.-R. Akbarzadeh-T. is with the Center of Excellence on Soft Computing
and Intelligent Information Processing (SCIIP) and Department of Electrical
(Control) Engineering, Ferdowsi University of Mashhad, Mashhad, Iran
(e-mail: akbarzadeh@ieee.org).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TNB.2012.2217382
1536-1241/$31.00 © 2012 IEEE
Authorized licensed use limited to: University of Szeged. Downloaded on May 22,2023 at 10:08:02 UTC from IEEE Xplore. Restrictions apply.
IEEE TRANSACTIONS ON NANOBIOSCIENCE, VOL. 11, NO.
selectively delivering drugs to a specific organ or tissue. Local
drug delivery by DENPs could reduce unwanted side effects
and thus increase the variety in choosing the type and dosage
of drug. DENPs have recently received significant attention as
a promising non-invasive approach to prevent and treat athero-
sclerosis in contrast to the existing methods such as global drug
delivery, angioplasty, and open surgery.
Hamzeh et al. [3] demonstrated that their designed targeted
DENPs, 80 nm in length and 30 nm in width, could successfully
enter the interior of the atherosclerotic plaques to deliver the
drugs and imaging agents in laboratory mice. The pioneering
works of Freitas on nanorobotics [14] proposed a desirable
medical nanorobot structure for targeted drug delivery. Suraj
and Reddy [4] proposed an algorithm for navigation of the
nanorobots towards the blood clot in the constricted arteries.
Hossain et al. [5], [6] mathematically modeled the transport of
coupled drug molecules and DENPs in coronary artery walls.
Martel [7] employed magnetic nanoparticles and a specific
MRI platform to design a swarm of synthetic microscale
nanorobots that are capable of traveling in the blood circulatory
network. Nakano et al. [8] formulated the dynamics of local
drug delivery by a nanoparticle-eluting stent and compared
its efficiency with the conventional stents. Nakano et al. [9]
theoretically considered the dynamic motion of the magnetic
nanoparticles in blood vessels and demonstrated that it could
be controlled by external magnetic fields. Dutta et al. [10] ex-
ploited the heat that is generated by iron-oxide nanoparticles in
presence of external magnetic field to melt the atherosclerotic
plaque. Furthermore, the nanorobot pioneer, Cavalcanti and his
colleagues [11] proposed a theoretical basis using chemical and
thermal gradients to navigate nanorobots towards the coronary
occlusion. Generally, these previous works are classified into
two categories: DENPs and nanorobots. The capabilities of a
nanorobot are much higher than a DENP, while manufacturing
of nanorobots is difficult by today’s nanotechnology.
This paper proposes a swarm of Proportional DENPs
(PDENPs) for swarm control of LDL level in the arterial wall
by exploiting piecewise-proportional control. The hardware
complexity of PDENPs aims to be simpler than nanorobots,
while the swarm architecture hopes to maintain overall perfor-
mance using swarm control. It is important to note that the goal
of the proposed PDENP is to quickly reduce the LDL level
in the interior of the arterial wall, not in the blood (lumen),
to prevent critical growth of atherosclerotic plaque. Thus,
PDENP is mainly designed for the patients with very high
LDL level ( [19]) who are corresponding to
ROWHANIMANESH AND AKBARZADEH-T.: CONTROL OF LOW-DENSITY LIPOPROTEIN CONCENTRATION IN THE ARTERIAL WALL
Fig. 1. Transverse section of the arterial wall [1].
the highest rates of cardiovascular disease events. In contrast
to the existing targeted DENPs that usually target the surface
proteins of atherosclerotic plaque, PDENP directly senses
the LDL level in the arterial wall, and hence it can diagnose
abnormal LDL accumulation before plaque formation. This is
a promising method to predict and prevent critical growth of
atherosclerotic plaques. PDENP is capable of distinguishing
between unhealthy and healthy arterial walls and therefore it
reduces the drug release rate and consequently the unwanted
side effects of drug in healthy tissues. We use an authentic
mathematical model of the arterial wall [1], [2] in computer
simulation to demonstrate the capabilities of PDENP.
The rest of the paper is organized as follows. In Section II,
mathematical modeling of the arterial wall such as the LDL,
DENP, and drug transport is briefly reviewed. The proposed
particle-based piecewise proportional control approach is ex-
plained in Section III. The general motivation and the structure
of a PDENP are also explained in this section. Simulation re-
sults are illustrated in Section IV. Finally, Section V concludes
the paper.
II. MATHEMATICAL MODELING OF THE ARTERIAL WALL
A. LDL Transport in the Arterial Wall
Fig. 1 schematically depicts the anatomical structure of the
arterial wall that is composed of six layers including glyco-
calyx, endothelium, intima, IEL (internal elastic lamina), media,
and adventitia [1]. There exist different mathematical models in
the literature for macromolecule transport in the arterial wall.
The present study utilizes the four-layer model introduced by
Yang and Vafai [1] in 2006 due to its generality and robustness.
This model has been also used in other researches [2], [5], [6],
[12]. The thickness of each wall layer is shown in Table I. In
this model, glycocalyx is ignored because of its small thickness
( 60 nm), and the effect of adventitia is modeled in boundary
conditions [1], [2].
For simplicity, the effect of LDL concentration on fluid ve-
locity is neglected, and fluid velocity is assumed to be con-
Authorized licensed use limited to: University of Szeged. Downloaded on May 22,2023 at 10:08:02 UTC from IEEE Xplore. Restrictions apply.
395
TABLE I
THICKNESS OF EACH LAYER IN THE ARTERIAL WALL [1]
Fig. 2. The boundary conditions used in the numerical simulation.
stant in all wall layers and equal to filtration velocity [6]. In this
study, the drug is defined as a chemical compound which tends
to combine with LDL, and the products of this reaction could
be excreted from the body through renal excretion. Without
loss of generality, the chemical reaction between drug and LDL
is presumed to be a second order reaction with coefficients of
and . Consequently, the governing equa-
tions of LDL concentration are [1], [2]
(1)
where is the layer number defined in Table I, the gradient,
the laplacian, the filtration velocity, and , ,
, and are the concentration, filtration reflection
coefficient, effective diffusivity and reaction coefficient (just
nonzero in media [1]), respectively, for LDL in the layer.
Fig. 2 and Table II represent the boundary conditions and the
physiological parameters that are used in the present simula-
tion. For the boundary conditions at the interfaces between wall
layers, the continuity of mass flux and concentration is consid-
ered [12].
B. PDENP and Drug Transport
In this paper, PDENP and drug molecule are assumed to
be spherical particles of 100 nm and 4 nm in diameter, re-
spectively. The previous experimental study by Hamzeh et al.
[3] demonstrated that DENPs of up to 80 nm in length and
30 nm in width could successfully enter the interior of the
arterial wall in laboratory mice. Moreover, Lu et al. [15] have
experimentally shown on laboratory mice that mesoporous
396
TABLE II
PHYSIOLOGICAL PARAMETERS FOR LDL [1]
silica nanoparticles, with spherical shape and 100–130 nm
in diameter, have rapid excretion from animal body through
urine and feces, suggesting a quick and complete clearance
of nanoparticles by the animal body [15]. As a result, the
assumed size for PDENPs in the present study is applicable.
An LDL molecule of 22 nm in diameter has approximately
got the proportion of 5 and 1/5 of the diameters of drug
molecule and PDENP, respectively. Therefore, the effective
diffusivity and filtration reflection coefficient for PDENP and
drug could be estimated as [2], [6]:
, ,
and . The boundary
conditions for PDENP and drug are displayed in Fig. 2. The
governing equations of PDENP and drug transport are
where is the mass of a drug molecule, and
the reaction coefficients of drug and PDENP with the environ-
ment, and and are the concentrations of drug and
PDENP in the layer, respectively. is the drug release
rate by each PDENP and manipulated with the
proposed controller.
III. PROPOSED CONTROL APPROACH
A. Motivations
Global drug delivery and targeted DENP are two of the most
important available non-invasive methods for treatment of
atherosclerosis. From control engineering perspective, global
drug delivery is open-loop control and targeted DENP is
semi-closed-loop control. In global drug delivery, the drug is
Authorized licensed use limited to: University of Szeged. Downloaded on May 22,2023 at 10:08:02 UTC from IEEE Xplore. Restrictions apply.
IEEE TRANSACTIONS ON NANOBIOSCIENCE, VOL. 11, NO. 4, DECEMBER 2012
directly released in the blood flow. Since there exist a huge
number of LDL molecules in the lumen and the drug tends
to combine with LDL, only small dosages of drug could be
administered to prevent drug toxicity. In this status, the whole
of the drug mass is approximately consumed in the lumen,
and only a trivial mass of drug is transported to the interior of
the arterial wall. More importantly, since no feedback is taken
from and , this method cannot distinguish between
unhealthy (abnormal) and healthy (normal) arterial walls, and
thus the drug is released blindly (open loop control). This
drawback could dangerously increase the unwanted side effects
on healthy tissues. Although treatment by global drug delivery
is economically much cheaper than treatment by DENPs, the
above-mentioned drawbacks reduce the effectiveness of this
method.
Targeted DENPs exploit protein bindings (or similar cases)
to target specific tissue for local drug delivery. These tar-
geted bindings play the role of feedback from the unhealthy
tissue. Most of the existing targeted DENPs target the surface
molecules of an inflamed plaque and this could sometimes be
a big limitation for this method. In many cases, the plaque
inflammation is trivial or in the early stages, but there is a very
high accumulation of LDL in the interior of the arterial wall.
In these cases, due to the low density of the desired proteins
(required for binding) on the plaque surface, the performance
, of targeted DENPs is not good. In general, since feedback is
not taken from LDL and drug, this method is open loop with
respect to LDL and drug. As a result, some of the drawbacks
of the global drug delivery are still available for this method,
but with less degrees.
B. Structure of a PDENP
A PDENP utilizes three simple units including sense, con-
troller and actuation, which enable it to continuously sense the
(2) human body and autonomously release the required drug if any
abnormality occurs. Fig. 3 depicts the structure of a PDENP.
The sense unit of PDENP includes two nanoscale molecular
concentration sensors to sense and from the en-
vironment. The controller unit is a simple memoryless piece-
(3) wise-linear mapping. The actuation unit of PDENP contains a
controllable drug pump (or a controllable drug valve connected
to drug payload) that releases the drug molecules in the aqueous
environment with a flow rate determined by the controller unit.
When a PDENP finishes its drug payload, it should be removed
and replaced by new PDENPs. For this reason, we consider an
artificial lysosome inside the PDENP that is conceptually sim-
ilar to a natural lysosome. When the PDENP is empty of drug,
the artificial lysosome releases its encapsulated enzymes inside
the PDENP, and consequently the PDENP is rapidly broken and
dissolved in the aqueous environment.
In this study, “biodegradability” is considered as the clear-
ance mechanism of PDENP. The task of artificial lysosome is
to break PDENP into small biocompatible molecules (similar to
the waste products of metabolism) which can be cleared from
living tissues through natural clearance mechanism of human
body. Generally, artificial lysosome takes action in two ways:
first, when the PDENP is empty of drug (released the whole of
ROWHANIMANESH AND AKBARZADEH-T.: CONTROL OF LOW-DENSITY LIPOPROTEIN CONCENTRATION IN THE ARTERIAL WALL 397

Fig. 3. The structure of a PDENP.

its encapsulated drug); second, after the lifetime of PDENP is

finished (even if PDENP is still full of drug). The second way
ensures that PDENPs are always cleared from human body es-
pecially from deep tissues like Media and lymph nodes.
It should be mentioned that in this paper, we consider a
nanoparticle as a system and our insight to PDENP is abstract
and mathematical. Thus, PDENP introduces an abstract model
for a nanoparticle, not a specific nanoparticle with certain ma-
terials and physical-chemical structure. But the materials that
are used in PDENP must guarantee biocompatibility especially
after biodegradation by artificial lysosome. Also, PDENP
should have a limited lifetime (above few days) such that artifi-
cial lysosome necessarily takes action after this lifetime (even
if PDENP is still full of drug). PDENP does not contain any
propulsion unit, while it exploits natural diffusion/advection
for locomotion such as macromolecules.
C. Piecewise Proportional Controller
A PDENP with piecewise-proportional controller is a simple
closed-loop controller that can address most of the above-men-
tioned problems. In contrast to the processing unit of the most
of the nanorobots that have been proposed in the literature
[16]–[18], the computational complexity of this controller is
lower. The dynamics of the proposed controller is as follows:
(4)
where and are the values of the LDL and drug con-
centrations in the environment and are measured by the sensors
of the PDENP, is the drug release rate, and
. , , and are the adjustable parameters of
this controller that are designed according to the desired perfor-
mance such as drug delivery rate.
PDENP is designed for patients with vey high LDL level
. To prevent releasing drug in lumen,
is set to zero for . Thus PDENPs do
not release any drug in lumen and the lower positions of En-
dothelium in which is above 160 . Also, to
prevent releasing drug in healthy arterial walls, is set
to zero for , where
Authorized licensed use limited to: University of Szeged. Downloaded on May 22,2023 at 10:08:02 UTC from IEEE Xplore. Restrictions apply.
Fig. 4. The control surface of the controller unit of each PDENP (Fig. 3) used
in the simulation.
the peak of the LDL concentration in healthy arterial wall is
. By this definition, in almost every point of
the Endothelium layer of the healthy arterial wall, in which
is between 10 and 100 , is
zero. Also, in the other layers of healthy arterial wall,
is very small and is propor-
tional to according to (4). As a result, the controller of
(4) have very low drug release rate in healthy arterial wall. This
means that PDENP can distinguish between healthy and un-
healthy arterial walls. To exploit the PDENPs that are trans-
ported to Intima, IEL and Media, is nonzero but small
for , corresponding to the LDL level in
these layers. Also, to increase the efficiency of drug release, a
feedback is taken from , and is linearly decreased
by increasing . Fig. 4 shows the control surface of this
controller.
There exist many research works in the literature such as
the pioneering works of Freitas and Cavalcanti in medical
nanorobotics [11], [16]–[18], that have just proposed theo-
retical basis and hypothetical structure for their nanorobots.
Some of those theoretical designs can not be implemented by
today’s nanotechnology yet. However, PDENP carries simpler
hardware architecture in order to be more reasonably realized
technologically. In recent years, many remarkable experimental
works have been performed on construction of computing unit,
sensors and actuators at nanoscale.
Some of the previous works on implementation of computing
(controller) unit at nanoscale include the realization of synthetic
genetic circuits on E. coli bacterium [20], building rewritable
digital data storage in live cells [21], construction of a DNA-
based computational platform for the assembly of a universal
398
set of logic gates, half-adder/half-subtractor system, multilay-
ered gate cascades, and parallel logic gate operations [22], im-
plementation of a NOR logic gate on an E. coli by manipulating
its chromosome, and combining few bacteria (NOR gates) using
quorum sensing (they called it “chemical wires”) to construct all
16 two-input Boolean logic gates in an aqueous environment
[23]. Also, various experimental studies have been done on the
construction of controllable nanometer-sized valves and pumps
that can be used for drug delivery such as designing a control-
lable nanometer-sized valve [24], construction of a reversible
molecular valve [25], and design of nano-pump using molec-
ular motor [26]. Similarly, there exit different works on building
nanoscale concentration sensors in the literature such as de-
signing a molecular concentration sensor based on the diffrac-
tion resonance mode of gold nanowire gratings [27], and con-
struction of polymeric nanowire chemical sensor [28]. Although
the size of components in most of these works is not less than
100 nm (the size of PDENP), they are promising enough to
demonstrate that much smaller nanoscale components will be
manufactured in near future to be applicable in PDENP.
IV. SIMULATION RESULTS
In this section, the proposed method is applied to control the
LDL level in the arterial wall. The mathematical model of the
previous section has been simulated numerically with the mesh
size of 100 nm and time step of 10 s in MATLAB [13]. It is
assumed that the proposed PDENPs are administrated to a pa-
tient with very high LDL level. Thus, the LDL concentration
in lumen, , is presumed to be 200 that is med-
ically known as “very high LDL level” [19]. We take the con-
centration of PDENP and drug in blood (lumen) as
and zero, respectively. Lu et al. [15] reported
that at concentrations below 100 , mesoporous silica
nanoparticles do not induce any cytotoxicity in a variety of cell
lines, and some growth inhibition was noted when the concen-
tration exceeds 200 . Thus, the value of in
the present study is applicable. The initial values of PDENP and
drug concentrations in the interior of the arterial wall are set to
be zero.
It should be mentioned that since we do not consider
a specific drug in this simulation study, some hypothet-
ical values are assumed for reaction coefficients between
drug and LDL, drug molecule mass, and other parameters
related to drug. In this simulation, , 24 hours) and PDENP concentration signal have a larger steady
, that is
100 times smaller than the mass of an LDL molecule, and
. The
maximum number of drug molecules to be encapsulated in a
PDENP is 15 000, and the mass of each PDENP is assumed
to be 60 000 times larger than . The maximum flow rate
of a drug pump is set to 1500 . This
value is relevant since in [14], the rate of the drug pump is
for a nanorobot of in diameter. It
is presumed that the fluid velocity is just nonzero towards
axis (Fig. 2), and equals to the filtration velocity, that is,
for the transmural pressure of
70 mmHg [1], [6]. Without loss of generality, in the present
simulation, a one-dimensional model is used in which signals
Authorized licensed use limited to: University of Szeged. Downloaded on May 22,2023 at 10:08:02 UTC from IEEE Xplore. Restrictions apply.
IEEE TRANSACTIONS ON NANOBIOSCIENCE, VOL. 11, NO. 4, DECEMBER 2012
are spatially dependent to axis [6]. It should be mentioned that
mass transport in the real arterial wall is a three-dimensional
phenomena and it is more complex than the one-dimensional
model of this simulation (due to lateral movement and diffu-
sivity of macromolecules in three-dimensional model). Hence,
for clinical applications, PDENPs should be simulated based
on the three-dimensional model of Section I. Fig. 4 illustrates
the control surface of the controller unit of each PDENP where
, and .
We consider the effect of PDENPs on the LDL level of the
interior of the arterial wall over 24 hours (one day) in two dis-
tinguishing situations: unhealthy (abnormal) arterial wall where
the peak of LDL concentration is 200 , and healthy ar-
terial wall where the peak of LDL concentration is 90 .
Figs. 5 and 6 represent the simulation results for these two cases,
respectively. In these figures, the plots of Parts a-c illustrate the
changes in LDL, PDENP and drug concentrations over time in
each wall layer. In these figures, “min,” “mean,” and “max”
stand for the minimum, average, and maximum value of the re-
lated signal over all points of the mentioned wall layer.
As Fig. 5(b) shows, at the first hours, because of the large
concentration gradient between lumen and Endothelium, large
amount of PDENPs are rapidly transported from lumen to
Endothelium and thus the PDENP concentration signal is
increasing. Due to transportation delay, after several minutes,
PDENPs are transported from Endothelium to the next wall
layers. Since at the first hours, the concentration of LDL is
high, PDENPs have high drug release rate according to (4).
This causes that most of PDENPs finish their encapsulated
drug and they are broken and dissolved in the aqueous envi-
ronment (biodegradability by artificial lysosome). As a result,
PDENP concentration signals decrease after few hours. But
since new PDENPs are transported from lumen to the arterial
wall, after about 6 hours, the input rate of new PDENPs and
the degradation rate of empty PDENPs reach an equilibrium
point and PDENP concentration signal reach its steady state
value in all wall layers. In contrast to diseased arterial wall,
Fig. 6(b) illustrates that since PDENP can distinguish between
healthy and unhealthy arterial walls, the drug release rate in
healthy arterial wall is much lower than unhealthy wall, and
thus PDENPs finish their encapsulated drug much later. This
causes that the input rate of new PDENPs and the degradation
rate of empty PDENPs reach an equilibrium point later (after
state value.
In following the above explanations, since there is no drug
and the LDL level is high in the arterial wall at the first hours, the
drug release rate by PDENPs is very large. Hence, as Fig. 5(c)
shows, the drug concentration is increased in the arterial wall.
But reaction between drug and LDL decreases the concentration
of drug and LDL in the arterial wall. Reduction of LDL level and
increase of drug concentration lead to lower rate of releasing
new drug molecules by PDENPs according to (4). Finally, after
about 6 hours, the release rate of new drug molecules and the
removal rate of drug (due to reaction) reach an equilibrium point
and drug concentration signal reaches its steady state value. A
similar process is occurred for healthy wall in Fig. 6(c), but with
longer settling time and very smaller steady state value.
ROWHANIMANESH AND AKBARZADEH-T.: CONTROL OF LOW-DENSITY LIPOPROTEIN CONCENTRATION IN THE ARTERIAL WALL 399

Fig. 5. The effect of PDENP on an unhealthy (abnormal) arterial wall (LDL level of 200 ) over one day. (a) LDL concentration vs. time. (b) PDENP
concentration vs. time. (c) Drug concentration vs. time. (d) Input concentration of PDENP from lumen to arterial wall at each time. (e) Average concentration of
the drug released from PDENPs into the arterial wall at each time. (f) Final LDL concentration profiles after one day in contrast to normal (desired) LDL level and
uncontrolled (without drug) LDL level. In this figure, “min,” “mean,” and “max” stand for the minimum, average, and maximum value of the related signal over
all points of the mentioned wall layer.

In the plots of Parts (d)–(e) in Figs. 5 and 6, the input con-
centration of PDENP from lumen to the arterial wall, and the
average concentration of the drug released from PDENPs in the
interior of the arterial wall are shown at each time. Regarding
these figures, almost all of the signals reach their steady state
values after 6 hours in the diseased case and after 12 hours in
the healthy case. This means that the proposed PDENP could
reduce the LDL level of the interior of the arterial wall to its
desired value and keep it stably at that level until the concentra-
tion of PDENP in blood is maintained at its administered value.
It should be noted that PDENP is not designed to reduce the
Authorized licensed use limited to: University of Szeged. Downloaded on May 22,2023 at 10:08:02 UTC from IEEE Xplore. Restrictions apply.
LDL level in blood flow, while it is designed to reduce the LDL
level in the interior of the arterial wall to prevent the critical
growth of atherosclerotic plaques. Hence, if the consumption of
PDENP is stopped by a patient with very high LDL level, since
LDL level is still very high in lumen (blood flow), the LDL level
will begin to increase in the interior of the arterial wall (because
of concentration gradient between lumen and Endothelium).
The plots of Part (f) depict the final (controlled) profile of
LDL concentration in the arterial wall at the end of 24 hours
and compare it with the desired (normal) and uncontrolled
(without drug) LDL levels. In the present simulation, the de-
400 IEEE TRANSACTIONS ON NANOBIOSCIENCE, VOL. 11, NO. 4, DECEMBER 2012

Fig. 6. The effect of PDENP on healthy (normal) arterial wall (LDL level of 90 ) over one day. The figure captions (a)–(f) are similar to Fig. 5.

sired (normal) LDL level is defined as the equilibrium point of respectively. According to this figure, the controlled LDL level
(1) when the concentration of LDL in lumen is 100 , is very close to the desired value in Intima, IEL, and Media.
corresponding to normal (optimal) LDL level, and is As it is expected, the controlled LDL level in Endothelium is
zero. The uncontrolled (without drug) LDL level is the initial higher than the desired value. This is because of the parameters
value of LDL concentration in this simulation and defined as of the controller such that it does not release any drug in the
the equilibrium point of (1) when is zero, and the con- lumen. Since the range of LDL concentration at the lower
centration of LDL in lumen is 200 for the unhealthy positions of Endothelium is close to lumen, the PDENP does
case (Fig. 5), and 90 for the healthy case (Fig. 6). not release any drug in this place and thus the LDL level is not
Fig. 5(f) demonstrates that although the LDL concentration significantly reduced.
in lumen is very high (200 ), the proposed PDENP Fig. 6(f) demonstrates that PDENP could successfully under-
could successfully reduce the LDL level in all layers of the stand that the arterial wall is healthy and it should not release
arterial wall with reduction rate drug in this tissue. In this case, LDL reduction rate is trivial and
of 17.9%, 45.7%, 0.09%, 3.49%, 3.83%, and 8.25% in Endothelium, Intima, IEL,
46.9%, and 61.3% in Endothelium, Intima, IEL, and Media, and Media, respectively. The area under the curves of Fig. 5(e)
Authorized licensed use limited to: University of Szeged. Downloaded on May 22,2023 at 10:08:02 UTC from IEEE Xplore. Restrictions apply.
ROWHANIMANESH AND AKBARZADEH-T.: CONTROL OF LOW-DENSITY LIPOPROTEIN CONCENTRATION IN THE ARTERIAL WALL
and Fig. 6(e) are 0.00390 and 0.00035 , [7] S. Martel, “Aggregates of synthetic microscale nanorobots versus
respectively. Comparing these two values shows that the mass
of the drug released in unhealthy (abnormal) arterial wall is
more than 11 times larger than the healthy wall. This confirms
the efficiency of the proposed PDENP in distinguishing between
unhealthy and healthy tissues which could significantly reduce
the unwanted side effects of drug.
V. CONCLUSION
This paper proposes a new control approach for treatment
of atherosclerosis that is one of the most common causes of
death in humans. Since high accumulation of LDL molecules
plays a major role in initiation and development of atheroscle-
rotic plaques, control of LDL within the arterial wall could be
a key approach to treat atherosclerosis. In this paper, Propor-
tional Drug-Encapsulated Nanoparticle (PDENP) is proposed
that utilizes a simple piecewise-proportional controller to re-
alize swarm feedback control of LDL level in the interior of the
arterial wall. The hardware complexity of the PDENP aims to be
simpler than nanorobots. Also, it has some advantages in con-
trast to the existing targeted DENPs that make it a promising ap-
proach to predict and prevent critical growth of atherosclerotic
plaques. PDENP is capable of distinguishing between unhealthy
and healthy arterial walls and therefore it reduces the drug re-
lease rate and consequently the unwanted side effects of drug
in healthy tissues. Simulation results on a well-known math-
ematical model of the arterial wall demonstrate that although
the LDL concentration in lumen is very high (200 ),
the proposed PDENPs successfully reduce the LDL level in
all layers of an unhealthy arterial wall with reduction rate of
17.9%, 45.7%, 46.9%, and 61.3% in Endothelium, Intima, IEL
and Media, respectively. Also, the mass of the released drug
by PDENPs in a healthy wall is 11 times less than its corre-
sponding value in the unhealthy wall. A further research in fol-
lowing this paper can be evaluating the performance of PDENPs
in three-dimensional model of the arterial wall by considering
the real-world parameters of a specific drug.
REFERENCES
[1] N. Yang and K. Vafai, “Modeling of low-density lipoprotein (LDL)
transport in the artery—Effects of hypertension,” Int. J. Heat Mass
Transf., vol. 49, pp. 850–867, 2006.
[2] L. Ai and K. Vafai, “A coupling model for macromolecule transport
in a stenosed arterial wall,” Int. J. Heat Mass Transf., vol. 49, pp.
1568–1591, 2006.
[3] J. Hamzah et al., “Specific penetration and accumulation of a homing
peptide within atherosclerotic plaques of apolipoprotein E-deficient
mice,” Proc. Natl. Acad. Sci. USA, vol. 108, no. 17, pp. 7154–7159,
2011.
[4] H. Suraj and V. B. Reddy, “QCA based navigation for nano robot for
the treatment of coronary artery disease,” in Proc. IEEE Int. Workshop
Med. Meas. Appl., 2011, pp. 131–136.
[5] S. S. Hossain, S. F. A. Hossainy, Y. Bazilevs, V. M. Calo, and T. J.
R. Hughes, “Mathematical modeling of coupled drug and drug-en-
capsulated nanoparticle transport in patient-specific coronary artery
walls,” in Computational Mechanics. New York: Springer, Aug.
2011, 10.1007/s00466-011-0633-2.
[6] S. S. Hossain, S. F. A. Hossainy, Y. Bazilevs, V. M. Calo, and T. J. R.
Hughes, “Mathematical modeling of coupled drug and drug-encapsu-
lated nanoparticle transport in patient-specific coronary artery walls,”
Institute for Computational Engineering and Sciences, University of
Texas at Austin, ICES Rep. 10-41, Oct. 2010.
Authorized licensed use limited to: University of Szeged. Downloaded on May 22,2023 at 10:08:02 UTC from IEEE Xplore. Restrictions apply.
401
swarms of computer-controlled flagellated bacterial robots for target
therapies through the human vascular network,” in Proc. 4th Int. Conf.
Quantum, Nano, Micro Technol., 2010, pp. 14–17.
[8] K. Nakano, K. Egashira, S. Masuda, K. Funakoshi, G. Zhao, S. Kimura,
T. Matoba, K. Sueishi, Y. Endo, Y. Kawashima, K. Hara, H. Tsujimoto,
R. Tominaga, and K. Sunagawa, “Formulation of nanoparticle-eluting
stents by a Cationic electrodeposition coating technology: Efficient
nano-drug delivery via bioabsorbable polymeric nanoparticle-eluting
stents in Porcine coronary arteries,” Cardiovascular Interventions, vol.
2, no. 4, pp. 277–283, 2009.
[9] M. Nakano, H. Matsuura, J. Dong-Ying, T. Kumazawa, S. Kimura, Y.
Uozumi, N. Tonohata, K. Koide, N. Noda, P. Bian, M. Akutsu, K. Ma-
suyama, and K. I. Makino, “Drug delivery system using nano-magnetic
fluid,” in Proc. 3rd Int. Conf. Innov. Comput. Inf. Control, 2008, p. 338.
[10] A. Dutta, S. S. Mohapatra, A. Sen, and S. Mukherjee, “Alternative for
by-pass surgery using iron-oxide nano-particles,” in Proc. Bio Micro
Nanosystems Conf., 2006, pp. 86–89.
[11] A. Cavalcanti, L. Rosen, B. Shirinzadeh, and M. Rosenfeld,
“Nanorobot for treatment of patients with artery occlusion,” in
Proc. Virtual Concept Conf., Cancun, Mexico, Nov. 2006.
[12] N. Yang and K. Vafai, “Low-density lipoprotein (LDL) transport in an
artery—A simplified analytical solution,” Int. J. Heat Mass Transf.,
vol. 51, pp. 497–505, 2008.
[13] B. Seibold, “A compact and fast Matlab code solving the incompress-
ible Navier-Stokes equations on rectangular domains,” Massachusetts
Institute of Technology, 2008 [Online]. Available: http://www-math.
mit.edu/cse/codes/mit18086_navierstokes.pdf
[14] R. Freitas, “Pharmacytes: An ideal vehicle for targeted drug delivery,”
J. Nanosci. Nanotechnol., vol. 6, pp. 2769–2775, 2006.
[15] J. Lu, M. Liong, Z. Li, J. I. Zink, and F. Tamanoi, “Biocompatibility,
biodistribution, and drug-delivery efficiency of Mesoporous silica
nanoparticles for cancer therapy in animals,” Small, vol. 6, no. 16, pp.
1794–1805, Aug. 2010.
[16] A. Cavalcanti and R. A. Freitas, Jr., “Nanorobotics control design: A
collective behavior approach for medicine,” IEEE Trans. NanoBiosci.,
vol. 4, no. 2, pp. 133–140, 2005.
[17] A. Cavalcanti, B. Shirinzadeh, T. Fukuda, and S. Ikeda, “Hardware
architecture for nanorobot application in cerebral Aneurysm,” in Proc.
7th IEEE Conf. Nanotechnol., Hong Kong, 2007, pp. 237–242.
[18] A. Cavalcanti, B. Shirinzadeh, M. Zhang, and L. C. Kretly, “Nanorobot
hardware architecture for medical defense,” Sensor, vol. 8, pp.
2932–2958, 2008.
[19] American Heart Association [Online]. Available: http://www.heart.org
[20] J. E. Norville, R. Derda, S. Gupta, K. A. Drinkwater, A. M. Belcher,
A. E. Leschziner, and T. F. Knight, Jr., “Introduction of customized in-
serts for streamlined assembly and optimization of BioBrick synthetic
genetic circuits,” J. Biol Eng., vol. 4, no. 1, p. 17, 2010.
[21] J. Bonnet, P. Subsoontorn, and D. Endy, “Rewritable digital data
storage in live cells via engineered control of recombination direction-
ality,” Proc. Natl. Acad. Sci. USA, vol. 109, no. 23, pp. 8884–8889,
2012.
[22] J. Elbaz, O. Lioubashevski, F. Wang, F. Remacle, R. D. Levine, and I.
Willner, “DNA computing circuits using libraries of DNAzyme sub-
units,” Nature Nanotechnol., vol. 5, pp. 417–422, 2010.
[23] A. Tamsir, J. J. Tabor, and C. A. Voigt, “Robust multicellular com-
puting using genetically encoded NOR gates and chemical ‘wires’,”
Nature, vol. 469, pp. 212–215, 2011.
[24] M. Yu, J. L. Falconer, T. J. Amundsen, M. Hong, and R. D. Noble, “A
controllable nanometer-sized valve,” Adv. Mater., vol. 19, no. 19, pp.
3032–3036, 2007.
[25] T. D. Nguyen, H.-R. Tseng, P. C. Celestre, A. H. Flood, Y. Liu, J. F.
Stoddart, and J. I. Zink, “A reversible molecular valve,” Proc. Natl.
Acad. Sci. USA, vol. 102, no. 29, pp. 10029–10034, 2005.
[26] D. J. Ahn, “Nano-Pump Using Molecular Motor,” U.S. Patent App.,
12/200,888, 2008.
[27] X. Zhang, F. Dou, and H. Liu, “Molecular concentration sensor based
on the diffraction resonance mode of gold nanowire gratings,” Nan-
otechnology, vol. 21, no. 33, p. 335501, 2010.
[28] H. Liu, J. Kameoka, D. A. Czaplewski, and H. G. Craighead, “Poly-
meric nanowire chemical sensor,” Nano Lett., vol. 4, no. 4, pp.
671–675, 2004.
Authors’ photographs and biographies not available at the time of publication.