In: Advances in Medicine and Biology.

Volume 52 ISBN: 978-1-62081-314-0

Editor: Leon V. Berhardt © 2012 Nova Science Publishers, Inc.

No part of this digital document may be reproduced, stored in a retrieval system or transmitted commercially
in any form or by any means. The publisher has taken reasonable care in the preparation of this digital
document, but makes no expressed or implied warranty of any kind and assumes no responsibility for any
errors or omissions. No liability is assumed for incidental or consequential damages in connection with or
arising out of information contained herein. This digital document is sold with the clear understanding that
the publisher is not engaged in rendering legal, medical or any other professional services.

Chapter VIII

Model Methodology for Analyzing

Peripheral Arterial Fractility from
Hemodynamic Data

Yoshihiro Nakamura1 and Shoichi Awa2

1
Department of Pediatrics, Graduate School of Medicine,
University of Tokyo, Japan
2
(Formely) Department of Pediatrics,
Kyorin University School of Medicine, Japan

Abstract
The fractal dimension provides us with biologically important meanings of the
branching pattern governing a tree-like structure and its space-occupying rules. Such
biological vascular structures have become well known in many recent reports as they are
simulated by fractal theory, where the fractal dimension was revealed to be 3 in the
branching pattern of arteries in organs and of bronchi in the lung. Murray’s law well
accounts for this dimension of 3 by the minimum work theory.
We deliberated the equation representing the relationship between the whole
peripheral arterial resistance and the total blood flow through it by combining the
fundamental fluid dynamics of Poiseuille with fractal theory. Our fractal hypothetical
stemming from the multifractal theory and arterial network were designed as an in-series
arrangement of parallel aligned multi asymmetric dichotomies, where the peripheral
resistive arteries start with a mean radius of 100 m and end at 10 m.
Letting x, R, and Q represent fractal dimension, total peripheral arterial resistance,
and total blood flow through the arterial network, respectively, and using body length
(BL) as the body scale parameter, we reached a proportionality of Q  (R/BL)-x/4. This
theoretical equation is applicable for determining peripheral arterial fractal dimensions in
vivo by only plotting in and processing the clinical data . The hemodynamic data of
patients with congenital heart defects of left-to-right (L-R) shunts were subjected to
analysis of the fractal dimensions of both pulmonary and systemic peripheral arteries
using our theory.
220 Yoshihiro Nakamura and Shoichi Awa

The fractal dimension of the peripheral systemic resistive arteries was estimated as
3.1 irrespective of the severity of the underlying PH. On the other hand, the dimension of
pulmonary arterial counterparts (xP) resulted in 2.2 in controls. xP gradually increased
from 2.2 to 3.0, with the severity of PH, until finally the pulmonary arterial pressure
reached systemic blood pressure level.
The PH dependency of xP in our model indicates the hampered development of intra-
acinar arteries and the more proximal involvement of radial reduction. These are
compatible with the pathological findings reported in many studies of L-R shunt defects
with secondary PH.
Our results suggest that the minimum work theory also works in vivo as the acquired
pathophysiological adjustment for reducing the preload of the left ventricle and
preserving the most important gas-exchanging function in the lung. This model provides
us with a powerful methodology for elucidating self-regulating design and mechanisms in
the living body from a theoretical standpoint.

Introduction:
Murray’s Law and the Relationship between Blood
Flow and Vessel Radius
C. D. Murray (1926) theoretically predicted the existence of an ‘optimality principle’ in
the relationship between the blood flow through a cylindrical vessel and the vessel radius
(Figure 1) [1]. The vascular resistance Rcyl of a cylindrical vessel is given in equation (1)
below as Hagen-Poiseuille’s equation when we hypothesize the vessel as a rigid one, letting
blood viscosity, radius, and length be , r, and l, respectively [1-7].

8l
Rcyl  (1)
r 4

Defining the volume blood flow rate per unit time, the metabolic rate of blood [1, 5] as
well as vessel wall volume [7] per unit time volume, and cost function representing the lost
energy per unit time to transport the blood flow through the vessel as q, K, and E,
respectively, he designed the equation as follows [1, 4-7]:
Figure 1
 8l 
E   4 q 2  Kr 2l . (2)
 r 

q
r

Figure 1. A single tubular vessel model. Blood flow (q) flows through the vessel whose radius and
length are represented by r and l, respectively. q is defined as volume blood flow per unit time.

Figure 2

E
 Figure 1
Model Methodology For Analyzing Peripheral Arterial Fractility … 221

After E was partially differentiated with respect to r, E/r =0 (Figure 2) resulted in [1, 4-
7]: l

q
r
 K
q   r 3 , where   . (3)
4 

Thus, he predicted that this relationship between q and r indicated by Eq. (3) must be
selected and adopted in the living body because this condition guarantees the minimum work
requirement, under which q is regulated and maintained proportionally to r cubed. The
validity of Murray’s law has been established by in vivo direct measurement of the
relationship between q and r in systemic microcirculations (Table 1) [7-9].
Figure 2

8 lq2/r4

Emin Kr2l

0 r
qr3

Figure 2. The graph of minimum work theory designed by Murray (1926) [1]. Emin represents the
minimum energy cost per unit time; , blood viscosity; q, r, and l, the blood flow, the vessel diameter,
and the vessel length in Figure 1, respectively. 8 lq2/r4 and Kr2l are presented in equation (2). Notice
that the point of Emin is given under the condition of q  r3 by partial differentiation of E with respect to
r. Figure 3

r1
q1
r0 q0

q2
r2

Figure 3. The branching model from the mother to two daughter vessels. q0, r0, and l0 represent the
mother’s blood flow rate, radius, and length, respectively; those of daughter 1 and 2 are presented with
suffixes 1 and 2, respectively.
222 Yoshihiro Nakamura and Shoichi Awa

Murray’s law can be interpreted as the ideal model of vessel branching [4-6, 10, 11].
When the mother vessel with her radius, r0, branches into two daughter ones with radii r1 and
r2 (r1 > r2), letting q0, q1, and q2 represent their blood flow through them, respectively, (Figure
3), the common K in Eq. (3) yields:

q0 = r03, q1 = r13, and q2 = r23. (4)

As the sum of blood flow through branching is preserved as q0 = q1+q2, Eq. (4) ensures:

r03 = r13 + r23. (5)

This is well known as the ideal relationship among radii in the branching of resistive
arteries. This radial relationship among three vessels at branching has been reported to hold
true in peripheral vessels by direct morphometric studies [5, 10].

Table 1. The radial exponent x in systemic and pulmonary arteries in the literature in
comparison with our model-derived x

Radius [m] x Method/Data Reference

Systemic
Rat, cremaster muscle 5*-50 2.73 IM/Fr [8], [9]
3.01  0.07
*
Rat, cremaster muscle 3 -54 IM/Fr [7]
Dog, arterioles 11-96 2.947 FA/Mm [6], [31]

Pig, coronary arteries 4.5*-25 3 FA/Mm [10]

Human, Cerebral arteries < 500 2.9  0.7 FA/Ag [53]
Our model, human 10-100 3.1  0.2 MA/Hd [22]
Pulmonary
Rat 13-800 2.15 FA/Mm [54]
Cat 12-508 2.35 FA/Mm [35], [55]
Human 30*-500 2.3  0.1 FA/Mm [19]
Our model, human 10-100 2.2 MA/Hd [22]
* represents not the mean radii of terminal arterioles but the minimum. IM represents in vivo
measurement; FA, fractal analysis; MA, model analysis; Fr, data of flow and radius; Mm,
morphometric data; Ag, angiographic data; Hd, hemodynamic data. Data is presented as a mean
with one standard deviation.

When repeated branching succeeds under this condition through the whole vascular tree,
the fractal dimension is also regarded as 3, which we will describe at the next section.
The shear-stress () on the internal surface of the vessel wall is measurable as 4q/r3 [9-
12] and is well known as an agent affecting and regulating vessel radius [13]. Because  of
three vessels is written by Eq. (4) as:
 Model Methodology For Analyzing Peripheral Arterial Fractility … 223

4 q0 4 q1 4 q2
   , (6)
 r03  r13  r23

Murray’s law also gives the background of the uniform shear-stress theory at the same
time [10-13].

Introduction of Fractal Theory to the Vascular

Tree
B. B. Mandelbrot (1983) reported the mathematical characteristics based on self-
similarity in geometric structures of seashore lines and rivers, or in morphology of tree
branching, where their self-similarity is described with fractal parameters and functions [14].
After his work, this concept of fractal theory has been developed and has become well known
worldwide. Research has also reported the presence of fractal structures in such living objects
as bronchial branching [15] and vascular networks [2, 7, 10, 16, 17]. When we suppose that
every k-th generation vessel with a mean radius of rk branches into N vessels on the average
of the k+1-th generation with radii of rk+1 through the vascular tree, letting k and N be a
natural number including 0 and an integer over 0, respectively, the fractal model is written by
using fractal dimension x as:

rkx = N rk+1x. (7)

Mandelbrot also called x the radial exponent in branching trees and vessels [14]. In the
fractal structure, vessels repeat branching succeedingly through generations of the vascular
tree obeying Eq. (7). When N =2 and x =3, Eq. (7) results in Eq. (5). Letting qk and qk+1
represent mean blood flow through a single vessel of the k-th and k+1-th generation,
respectively, preserved blood flow is written as:

qk = N qk+1. (8)

Eqs. (7) and (8) yield the next equation similar to Eq. (4) with the fractal dimension x
as[7, 9, 16, 18, 19]:

q =  rx . (9)

Meanwhile, morphometric analysis reported x 2 in Eq. (7) at the branching of the

proximal systemic large arteries originating directly from the aorta [20, 21] and the normal
pulmonary arterial tree (Table 1) [17, 19, 21], and these bifurcating vascular structures have
been accounted for by the square law for preserving cross-sectional area through branching
[5, 20, 21]. The fractal dimension of bifurcating peripheral resistive arteries, on the other
hand, were reported to be x =3 in morphometric studies [5, 10], as well as featuring flow-
radius measurements in vivo as described above [7-9] (Table 1). x =3 in organ vasculatures
were also theoretically accounted for by the uniform pressure distribution and optimal space-
224 Yoshihiro Nakamura and Shoichi Awa

filling of fractal dimensions [4] as well as by Murray’s law. The heterogeneity of x from 2 to
3 was also reported in whole pulmonary arterial trees of dogs [3, 12, 18]. Actually, the
multifractility [23-25] exists in the vasculature of mammals [23, 24], where the fractal
characteristics may vary at different levels of radii.

1. A Fractal Model for Translating Flow-Radius

Relationships of a Single Vessel into
Flow-Resistance One throughout the
Whole Vascular Tree
A new mathematical model reported by Nakamura et al. (2011) was designed with fractal
theory and Eq. (9) by translating the relationship between q and r of a single vessel into that
between the total blood flow (Q) and the whole peripheral arterial resistance (R) in an arterial
network [22]. This model can be applied to the fractal analysis of resistive arteries and
peripheral arterial systems with the hemodynamic data clinically accessible in a routine cath
study. We defined the unit resistive arterial tree as an asymmetric branching fractal tree
whose proximal generation number is given as the 0-th and most peripheral one as the n-th,
letting n be a natural number including 0. Letting k be an integer as 0 k  n, when the k-th
generation vessels whose means of radius, vessel length, and number of vessels are given as
rk, lk, and Nk, respectively. Large arteries (r >100 m) which function as an elastic reservoir
or a Windkessel [26] share 10% of the mean aortic pressure (PAo), while 60% of PAo is lost in
small resistive arteries with r 100 m [2]. First, we constrained the mean radius of the initial
starting 0-th resistive artery as r0 =100 m. Parallel N0 unit resistive arterial trees are regarded
as scatteringly
Figure 4connected to large arteries in-series in our model (Figure 4A).

A B
generation
0 1 2 ...……k k+1…n-1 n

qk+1
qk qn
q2 qn
q1
qn
q0
. . . . . . . . . . . . . .....

qk
................

qk+1
...

...

qk qk+1

qk qn
qk+1

Figure 4. A. A schematic diagram of either the pulmonary or systemic arterial tree. Unit resistive
arterial trees, scatted and connected in parallel to the central compliant artery, are represented by
circles. B. A schematic diagram of a unit peripheral resistive arterial tree under the condition of N =2 in
Eq. (10),Figure
which5starts from the 0-th generation, bifurcating repeatedly in series until the terminal n-th.
The last n-th generation vessels are distally connected to capillaries. qk, mean blood flow through the k-
th generation vessels,5 where arrows () indicate the direction of blood flow; k, natural number from 1
to n. Figures were revised
4
from the original article [22].
[cP]

0
1 10 102 103 104
radius [m]
 Model Methodology For Analyzing Peripheral Arterial Fractility … 225

Figure 5. A schematic diagram suggesting the landscape around terminal arterioles. Scattered ovals
represent red blood cells (RBC). The mean radius (rn) of terminal arterioles is considered to be
determined by the scale of RBC. So, rn turns common among different species of mammals because the
size of mammalian RBC stays almost same among them. qn and ln represent mean blood flow and
vessel length of terminal arterioles, respectively.

The radius of the n-th terminal arteriole is reported to be around 7-10 m in rat’s
systemic [8, 9], canine systemic [27], canine pulmonary [3], and human pulmonary [28]
arteries, because radii of terminal arterioles are similar in spite of different mammalian
species (Table 1, Figure 5. This data suggests that the scale of red blood cells, whose diameter
are around 7 m, is critical in determining the radius of terminal arterioles as well as that of
capillaries (5 m) [2, 29]. So, we constrained the mean radius of the last n-th arterioles as rn
=10 m, where Nn terminal arterioles are connected to capillaries (Figure 4B). Capillaries are
not included into our model because their network structure is not a simple tree but a complex
sinusoid [30].
We defined the fractal rule from the k-th generation to the k+1-th generation as:

rk 1 l N k 1
 ,   k 1 , and N . (10)
rk lk Nk

where , , and N are all real numbers representing ratios of radius,
length, and branch numbers, respectively. Then, Eq. (10) results in:

rn = n r0, ln = n l0, and Nn = Nn N0. (11)

Because vessels of the same generation are arranged in parallel with each other, the total
arterial resistance of k-th generation vessels (Rk) is given with Eqs. (1), (10), and (11) as:

k kn
8lk    8l0    8ln
Rk        . (12)
N krk N   r0 N 
4
 rn

As the whole vascular tree in question is made up of each generation placed in series as
shown in Figure 4B, the total resistance of peripheral resistive arteries (R) is given as:

 226 Yoshihiro Nakamura and Shoichi Awa qk

n
R   Rk . (13)
k 0

The radius-dependent blood viscosity known as the Fähraeus-Lindqvist effect [2, 31] is
effective in this resistive arterial area, where r 100 mm. So we use the equation reported by
Haynes [32] as:


(r)  , (14)
(1 rt /r)
Figure 5
where  and rt is 4.0 centipoise (cP) and 4.29 m, respectively (Figure 6) [32, 33].

 5

4
[cP]

0
1 10 102 103 104
radius [m]
Figure 6. The simulation curve of Fähraeus-Lindqvist effect [2, 31] reported as Haynes’ equation [32,
33].  represents the blood viscosity given as centipoise (cP); r, radius.

We define the ratio of rn to rt as:

rt
 . (15)
rn

Eqs. (12), (14), and (15) are incorporated into Eq. (13) as:

n
8lk 8ln
R   G ( n) , (16a)
k  0 N k  (1  rt / rk ) rk N n rn4
2 4

n
 kn
where G(n)   N kn 2(kn )( kn  )2 . (16b)
k 0


 Model Methodology For Analyzing Peripheral Arterial Fractility … 227

Letting whole blood flow through this arterial network be Q, Eqs. (8) and (9) yield:

Q = Nn rnx. (17)

Eq. (17) is then used to Eq. (16a) as:

8   ln
R  Q  4 / x 4 / x N n4 / x G ( n) . (18)
 Nn

Eq. (18) gives Q as:

 8 l
4/ x

Q   N n1 4 / x   n G(n)  R4 / x . (19)
  

We designed the vessel length to be proportional to some body scale parameter M for
standardization. So, we considered that ln = dn·M and Eq. (19) resulted in:

4 / x
 8 d
4/ x
 R
Q   N n1 4 / x   n G(n)    . (20)
   M 

Bilaterally logarithmic expression of Eq. (20) yields:

R
ln Q  A ln    B , (21a)
M 


 1 4 / x  8 d n  
4/ x

where and B  ln  N n  G ( n)   . (21b)

     

As shown in the next section, this model was actually applied to clinical hemodynamic
data in order to determine fractal parameters. For our model analysis, we subjected the
clinical hemodynamic data of congenital left-to-right (L-R) shunt defects. Ventricular septal
defects (VSD), atrial septal defect (ASD), and patent ductus arteriosus (PDA) are the 3 most
common L-R shunt defects at either the ventricular, atrial, or aortic level, respectively. In vivo
obtained cath data of 213 patients with L-R shunts prior to surgical operation were available
for analysis.
Fractal model parameters and vessel numbers of x, n, , , N0, and Nn were estimated by
solving an inverse problem of Eqs. (11), (16a), and (21b) with some appropriate constraints.
As a control group, we selected patients whose pulmonary arterial pressure was normal and
whose L-R shunt was estimated at 0% because of a tiny defect and the limitation of
measurement [22]. Hereafter, suffixes S and P stand for systemic and pulmonary,
respectively.
 228 Yoshihiro Nakamura and Shoichi Awa

2. Systemic Circulation and the Importance of

qn
Body Length in ther
New Model
ln n

Figure 7B schematically indicates the longitudinal distribution of systemic pressure loss.

While arteries lose 70% of mean aortic pressure, capillaries and veins lose 30% of it. As
described above, the ratio of arterial pressure loss in proximal large arteries (r >100 m) and
peripheral resistive arteries (r 100 m) is 1:6 [2]. For simplicity we suppose N =2 in Eqs.
(10) and (11) for this model formulation because the arterial branching pattern has long been
Figure 7
treated as a bifurcation in model analysis of systemic circulation [21, 34] (Figure 4B and 5).
The detail of the method to extract the peripheral systemic arterial resistance RS was
described in our article [22].

A B
Pulmonary circulation Systemic circulation

Proximal large arteries

100 m < radius
PmPA - PLA

Peripheral resistive

PAo
arteries
10 m ≤ radius ≤ 100 m

Capillaries & veins

Grade of PH due to L-R shunt
Body Scale
Normal

Figure 7. These schemas illustrating the relative longitudinal distribution of three segmental pressure
gradients in pulmonary (A) or systemic (B) circulation in relation to pulmonary hypertension (PH) and
systemic blood pressure, respectively. PmPA PLA, and PAo indicate mean pressures of the main
pulmonary artery, left atrium, and aorta, respectively. These three segments are proximal large arteries
of radius >100 m, peripheral resistive arteries of radius 10-100 m, and capillaries and veins. We
applied 1:6 as the ratio between pressure drops through the proximal pulmonary arteries and through
the peripheral pulmonary arteries [2]. The pressure drop through capillaries and veins does not change
even in PH due to chronic L-R shunt [26]. Figure 7A were revised from the original article [22].

2.1. Body Length and Actual Methods for Estimating x

Relationships between systemic blood flow (QS) and peripheral systemic arterial
resistance (RS) was analyzed. First, the simple plot of lnQS vs. lnRS of all patients is presented
in Figure 8A. lnQS and lnRS turned out to be linearly arranged. As standardized by body
surface area (BSA), ln(QS/BSA) vs. ln(RS·BSA) are presented in Figure 8B for reference.
 Model Methodology For Analyzing Peripheral Arterial Fractility … 229

Figure 8. A. Simple plot of lnQS vs. lnRS. B. Plot of lnQS/BSA vs. lnRSBSA. C. Plot of lnQS vs.
lnRS/BSA. D. Plot of lnQs vs. lnRS/BW, QS, RS, BSA, and BW stand for systemic blood flow [L/min],
peripheral systemic arterial resistance [mmHg/L/min], body surface area [m2], and body weight [kg].
The regression line was fitted by the least square method. r stands for correlation coefficient.

Then, three different body scale parameters for M were introduced to standardize RS as
indicated in Eq. (21a). We compared the regression slope and correlation coefficient for each
M, e.g. BSA, body weight (BW), and body length (BL) as shown in Figures 8C, 8D, and 9,
respectively. BL was defined as the distance from the top to heel. When BSA and BW were
applied to Eq. (21a) as M, the slope (= A) of the regression line resulted in -0.637 (Figure 8C)
and -0.549 (Figure 8D), respectively, with close negative correlations (r =-0.985 and -0.981,
respectively).
Finally, the plot applying BL to Eq. (21a) was presented in Figure 9. lnQS vs. ln(RS/BL)
plots exhibited A =-0.721 (r =-0.983) in controls (Figure 9A), and A =-0.776 (r =0.987) in all
patients under the study, respectively (Figure 9B).
230 Yoshihiro Nakamura and Shoichi Awa

Figure 9. lnQS vs. ln(RS/BL). A. Plot of controls. B. Plot of all patients. QS, RS, and BL stand for
systemic blood flow [L/min], peripheral systemic arterial resistance [mmHg/L/min], and body length
[m], respectively. The regression line was fitted by the least square method. r stands for correlation
coefficient. Figure 9B was revised from the original article [22].
The regression line with the least scatter was found in Figure 9B, where BL was used as the most
suitable body scale parameter. Furthermore, with Eq. (21b), these values of A resulted in x =2.9-3.1,
which fluently accounted for Murray’s law (Table 1). On the other hand, BSA and BW did not represent
Murray’s law from their regression slopes with a result of x =2.5 and x =2.2, respectively.

Therefore, introduction of BL is appropriate to adjust to different vessel lengths and meet

Murray’s law in systemic resistive arteries. Thus, M in Eq. (21a) was rewritten with BL and
completed as:

 R 
ln Q  A ln  B. (22)
 BL 

We attempted to apply Eq. (22) to pulmonary circulation in section 3 ahead.

2.2. Methods for Estimating n, , , N0, and Nn

Because Q should be preserved throughout the arterial system, Eq. (11) is as follows:

Q  N0 r0x  Nn rnx  2n N0 rnx . (23)

with N =2, r0 =100 m, and rn =10 m. Eq. (23) gives mean n as x (ln10)/(ln2), and mean n
is estimated as 10.3 from x =3.1.  and  are also computed from estimated means n, r0 =100
m, and rn =10 m with an estimated l0, ln, and constrained  from morphometric data in past
reports. N0, and Nn are estimated with those results of , , and n by Eqs. (11), (16a), and
(16b). We take n as the nearest integer of mean n in calculating Eq. (16b). West et al. reported
 = = N -1/3 from the fractal model analysis based on the space-filling principle [30]. Because
N is equal to 2 in this model, it follows that =  =0.794. Our results of  and  in systemic
 Model Methodology For Analyzing Peripheral Arterial Fractility … 231

arteries were 0.80 and 0.83 (Figure 10B), which were in agreement with 0.794. Estimated N0
and Nn were also in agreement with corresponding counterparts of canine morphometric data
[36] as the best available substitution for direct comparison because no pertinent
morphometric data in humans is available in literature (Table 2).

Table 2. Comparison of our model-derived estimates in our control group for

generation, order, and number of vessels with pertient morphometric data in the
literature

Radius Generation Order Number of vessels Reference

[m]
Systemic
Dog, 20 kg, whole arterial tree 75 1.1  105 [36]
4* 2.7  109*
Dog, mesenteric artery 96 (3) (4‡) 6.1  102 [6], [27]
11 (6) (1‡) 1.3  106
Our model, human, 15.7 100 0 11 4.0 (0.8-11.6)  104 [22]
(5.4-66.6) kg
10 10 1 4.9 (1.0-14.2)  107
Pulmonary
Human, 112  13 kg 110  15 0 6 2.85  105 [28]
10  1 5 1 5.12  107
Human, 56-years-old Male 112 0 7 5.81  104 [56]
10.5 5 2 2.03  107
Human, 32-years-old Female 112 0 7 5.23  104 [57]
10.5 5 2 1.27  107
Cat, 3.9-5.9 kg 96 0 5 2.93  103 [55]
12 4 1 3.00  105
Our model, human, 15.7 100 0 (0§) 8 (5§) 3.6 (0.6-19.0)  105 [22]
(5.4-66.6) kg
10 7 (4§) 1 (1§) 5.5 (0.9-29.6)  107
* represents data of capillary. ‡ represents ‘rank’ in original paper. The number of rank is not
exactlycorresponding to order.§ represents the result in case of N =3.58 (Yen et al. 1984 [35]); N,
the ratio of number of vessels in Eq. (10). The ‘order’ is the rule of reversely numbering arteries
from the most peripheral terminal arteriole backward to the most proximal artery [28, 55-57]. The
number of vessels in our model is presented as median with range.

3. Application of This Model to Normal Pulmonary

Circulation
The distribution ratio in transpulmonary pressure loss is reported to be 50:50 between the
pulmonary arterial system and the region of pulmonary capillaries and veins [26, 37-40]. We
assumed that the proximal pulmonary arteries (r >0.1 m) and the peripheral ones (r 0.1
m) share the total pulmonary arterial pressure loss at a ratio of 1:6 (Figure 7A), respectively,
both on the basis of the canine in vitro data directly measured by capillary micropuncture
techniques [40] and on the analogy of the distribution in systemic arteries [2]. While the
branching ratio N reported by morphometric studies in the pulmonary arterial tree varied from
1.73 in 2-month lambs [17] to 3.58 in cats [35]; generally N in pulmonary artery has also been
treated as equal to 2 in model analysis [3, 16]. So, N in this model is basically treated as 2 for
simplicity in pulmonary arteries as well as systemic ones in this model (Figure 4B and 5).
232 Yoshihiro Nakamura and Shoichi Awa

Figure 10A presents the plot of lnQP vs. lnRP/BL in our control group. The regression slope A
yielded x =2.2 which is in agreement with the data reported by past morphometric studies
(Table 1).  and  in normal pulmonary circulation were estimated as 0.73 and 0.80,
respectively. Our  was in agreement with the model analysis of Dawson et al. under N =2,
where their  resulted in 0.77 [3]. Yen et al. reported that N, , and  were 3.58, 0.58, and
0.55, respectively, in the cat’s pulmonary arterial tree [35]. When N is assumed to be 3.58 in
our model,  and  are given as 0.57 and 0.64, respectively, which are consistent with the
results of Yen et al. Estimated values of n, N0, and Nn of our model in normal pulmonary
circulation were compared with past morphometric data and was found to be in agreement
with the given data (Table 2). We partially verified our supposed ratio of 1:6 with our own
pressure gradient data at the first branching from the main pulmonary artery in our article
[22]. But this ratio does not change the regression slope (A). So, basic parameters of x, , and
 are not influenced by the ratio.

Figure 10. lnQP vs. ln(RP/BL). A. Plot of controls. B. Plot of all subjects categorized to 7 groups
including controls in terms of severity of pulmonary hypertension induced by left-to-right shunt. QP, RP,
and BL stand for pulmonary blood flow [L/min], peripheral pulmonary arterial resistance
[mmHg/L/min], and body length [m], respectively. PmPA and PLA indicate mean pressures of the main
pulmonary artery and left atrium, respectively. QP/QS stands for the ratio of pulmonary vs. systemic
blood flows; r, the correlation coefficient. The regression line was fitted by the least. Figure 10B was
revised from the original article [22].

4. Application to the Pulmonary Circulation with

L-R Shunts
The pressure loss of pulmonary capillaries and veins in dogs with experimentally
produced L-R shunts and secondary PH remains the same with controls [26]. On the basis of
this report, we assumed that the peripheral pulmonary pressure loss shares the same ratio of
1:6. Please refer to the details in our article for further explanation [22]. Figure 7A shows the
supposed distribution of transpulmonary pressure loss for both controls and secondary
pulmonary hypertension (PH) induced by L-R shunts.
 Figure 11 Model Methodology For Analyzing Peripheral Arterial Fractility … 233

A B

pulmonary
4.0 1.00  pulmonary
systemic P= 0.017lnPP + 0.792
xP = 0.254lnPP + 2.040
 systemic
r = 0.903
r = 0.963 0.90
3.0
0.80

,
x

 pulmonary
0.70  systemic
2.0  P = 0.022lnPP + 0.714
r = 0.904
0.60

1.0 0.50
0 10 20 30 40 50 0 10 20 30 40 50

P [mmHg] P [mmHg]

Figure 11. A. The change of x in relation to P. B. Those of  and  in relation to P. The mean with 
one standard deviation is presented as the circle with horizontal outliers, respectively. Dashed curves
represent the change of xP, P, or P to PP by fitting a logarithmic function with the least-square
method. r represents the correlation coefficient. P means the peripheral arterial pressure drop; suffix P,
pulmonary. Systemic P is estimated as 60% of mean aortic pressure, while pulmonary one as 60% of a
half of transpulmonary pressure gradient, which is given by mean pulmonary arterial pressure minus
mean left arterial one. Figure 11A was revised from the original article [22].

Figure 10B presents the plot of lnQP vs. lnRP/BL in controls and 6 groups into which we
categorized patients according to the severity of PH due to L-R shunts. For details of patient
categorization, refer to Figure 10B and our article [22]. Figure 10B shows the gradually
decreasing slope of regression lines in relation to the severity of PH. The changes of slope
resulted in the increasing of x from 2.2 to 3.0 (Figure 11A), which alsoended in n from 7 to 10
(Figure 12A and B). Gholishi et al. (2007) reported a mean value of x to be 1.671 by
morphometry in lambs with artificially placed L-R shunts in their fetal life. However, their
results reflected the dimension of more proximal arteries than ours [17], while their Figure 2B
clearly exhibited the increase of x coming up to 3.0 in the most peripheral pulmonary
arteriolar region. Nichols and O’Rourke described that cast technique involves errors that are
difficult to allow for [2]. The high injection pressure of resin [17, 41], a linear shrinkage of at
least 10 percent on setting [2], the non-physiological temperature of gelatin-barium
suspension at 60C [17, 41], and high viscosity of resin (methlmethacrylate plastic) [17]
might not be suitable for estimating the micro-structure at the pulmonary terminal arteriolar
level [41]. Furthermore, as the actual in vivo measurement of hemodynamic data in peripheral
pulmonary arteries under the voluntary respiration is unfeasible, which is also the case in in
vitro ones under the artificial positive-pressure ventilation, the model analysis with in vivo
hemodynamic data of routine catheterization under the voluntary respiration is important.
Changes of estimated  and  in terms of the severity of PH were presented in Figure
11B. These changes of n, , and  indicated slenderization and elongation of cumulative
lengths of resistive arterial trees as well as altered radii and lengths of each generation (Figure
12A and B). The change of the whole vessel number (NW-P) of peripheral pulmonary arteries
(10 r 100 m) in relation to PH is presented in Figure 13, where NW-P was standardized
with BSA. NW-P represents the total number of intra-acinar arteries of the lung [42].
Figure 12

234 A Yoshihiro Nakamura and Shoichi Awa

0 1 2 3 4 5 6 7
A mean radius of 100 m shifted proximally by 3 generations in the severest PH group
200 m recovered from PH immediately or at most several months
(Figure 12B). Because all patients 20 m
after surgical operations, their PH proved reversible and could be accounted for by
Figure 12 3 mm
vasoconstriction [43-45] due only to transient medial hypertrophy [41, 46, 47].

BA
0 10 21 23 3 4 4 5
5 6 67 7 8 9 10

200 m 20 m
200 m 20 m
3 mm

4 mm
B
0 1 2 3 4 5 6 7 8 9 10

200 m 20 m

4 mm

Figure 12. The schematic illustration indicates vessel diameter and length of each generation of a unit
resistive pulmonary arterial tree for either control (A) and the group with severest pulmonary
hypertension (B) in our data for BL =1.2 m; BL, body length.

Figure 13
The decrement of NW-P/BSA in accordance with the severity of PH (Figure 13) suggested
hampered development of intra-acinar arteries due to L-R shunts. Analytic results with our
model were as a whole in agreement with human pathological findings in patients with
Figure 13
secondary PH due to L-R shunts [45, 46].

109

109
NW-P/BSA = 8.804109 PmPA-1.541
NW-P/BSA = 8.804109 PmPA-1.541

108
/BSA

108
NW-P/BSA
N
W-P

7 7
1010

106
106
0 20 40 60 80

0 P20
mPA [mmHg] 40 60 80
PmPA [mmHg]

Figure 13. Plot of NW-P/BSA vs. PmPA. NW-P stands for whole number of peripheral pulmonary resistive
arteries which corresponds to that of intra-acinar arteries [42]; PmPA, mean pressure of main pulmonary
artery [mmHg]. BSA, body surface area [m2]. Lines are exponential regression curves; r, correlation
coefficient.
 Model Methodology For Analyzing Peripheral Arterial Fractility … 235

However, these pathophysiological changes producing PH are desirable in some respects

with certain benefits. The vasoconstriction and suppressed development of intra-acinar
arteries might play an important part in impeding further high flow, pulmonary edema, and
capillary damages by restricting the increment of QP because pulmonary capillaries are
crucial and delicate regions in the lung for exchanging gases. Furthermore, these adjustments
to PH in peripheral pulmonary arteries lead to a decreasing of the preload of the left ventricle,
which is the most important pump for the living body. The systemic circulation must remain
stable in spite of increased pressure load to the right ventricle and peripheral pulmonary
arteries. It is needless to say that the long-term pressure load to peripheral pulmonary arteries
causes irreversible lesions of intimal thickening [46, 48-50], thrombosis, and vasculitis [26,
46, 50-52], which hamper the actual long-term prognosis of life. But without these
adjustments of pulmonary resistive arteries, the short-term survival would be jeopardized
because of lethal pulmonary edema and severe congestive heart failure.

Conclusion
We developed a new mathematical model to translate blood flow and radius in a single
vessel to whole blood flow and total peripheral arterial resistance in the whole arterial system
based on fundamental hydrodynamic laws and multifractal theory. This model was applicable
to peripheral resistive arteries in both systemic and pulmonary circulation as a method for
determining the fractal dimension, related fractal parameters, and number of the vessels.
This model is designed on the basis of the hypothesis that the mean length of the vessel is
proportional to its body length. The hypothesis cannot be easily verified with our data alone
or past morphometric data in pertinent literature at this time. Nevertheless, our design and
central concept of the model is at least indirectly supported by the result of x =3 in peripheral
systemic arteries because the validity of Murray’s law (x =3) is established in the same
region. Furthermore, our hypothesis is indirectly supported by the results of fractal parameters
and the number of vessels in accordance with past model analysises and morphometric data in
both systemic and normal pulmonary arteries. We also found this model has more
applicability to the structural and quantitative estimation of peripheral arteries under the
pathophysiological circulation. Our model was applied to the analysis of the
pathophysiological changes of peripheral pulmonary arteries under the condition of secondary
hypertension due to enhanced pulmonary blood flow. Results from our model analysis were
in agreement with reported pathological findings in L-R shunts and PH, namely, the
hampered development of intra-acinar arteries and vasoconstriction. Our model accounted for
functional, structural, and quantitative findings reported by past clinical pathophysiology
research.
The model analysis with clinical data is important and effective as a method for
quantitatively estimating such structural parameters to define the most peripheral arterial
region, which is difficult to accomplish in vivo. We hope that this model might shed more
light on the pathophysiology of human circulation from a theoretical standpoint.
236 Yoshihiro Nakamura and Shoichi Awa

References
[1] Murray, C. D. (1926). The physiological principle of minimum work, I: The vascular
system and the cost of blood volume. Proc. Natl. Acad. Sci. USA, 12, 207-214.
[2] Nichols, W. W. and O’Rourke, M. F. (2005). McDonald’s Blood Flow in Arteries:
Theoretical, Experimental and Clinical Principles (fifth edition). London, UK: Hodder
Arnold.
[3] Dawson, C. A., Krenz, G. S., Karau, K. L., Haworth, S. T., Hanger, C. C., and Linehan,
J. H. (1999). Structure-function relationships in the pulmonary arterial tree. J. Appl.
Physiol., 86, 569-583.
[4] Gafiychuk, V. V. and Lubashevsky, I. A. (2001). On the principles of the vascular
network branching. J. Theor. Biol., 212, 1-9.
[5] LaBarbera, M. (1995). The design of fluid transport system: a comparative perspective.
In: J. A. Bevan, G. Kaley, and G. M. Rubanyi, (Eds.), Flow-Dependent Regulation of
Vascular Function (pp. 3-27). New York, USA: Oxford University Press.
[6] Sherman, T. F. (1981). On concerning large vessels to small. J. Gen. Physiol., 78, 431-
453.
[7] Mayrovitz, H. N. and Roy, J. (1983). Microvascular blood flow: evidence indicating a
cubic dependence on arteriolar diameter. Am. J. Physiol., 245, H1031-H1038.
[8] House, S .D. and Lipowsky, H. H. (1987). Microvascular hematocrit and red cell flux in
rat cremaster muscle. Am. J. Physiol., 252, H211-H222.
[9] Lipowsky, H. H. (1995). Shear stress in the circulation. In: J. A. Bevan, G. Kaley, and
G. M. Rubanyi, (Eds.), Flow-Dependent Regulation of Vascular Function (pp. 28-45).
New York, USA: Oxford University Press.
[10] Kassab, G. S. and Fung, Y. C. (1995). The pattern of coronary arteriolar bifurcations
and the uniform shear stress hypothesis. Ann. Biomed. Eng., 23, 13-20.
[11] Schreiner, W., Karch, R., Neumann, M., Neumann, F., Roedler, S. M., and Heinze G.
(2003). Heterogeneous perfusion is a consequence of uniform shear stress in optimized
arterial tree models. J. Theor. Biol., 220, 285-301.
[12] Karau, K. L., Krenz, G. S., and Dawson, C. A. (2001). Branching exponent
heterogeneity and wall shear stress distribution in vascular trees. Am. J. Physiol. Heart
Circ. Physiol., 280, H1256-H1263.
[13] Wasserman, S. M. and Topper, J. N. (2004). Adaptation of the endothelium to fluid
flow: in vitro analyses of gene expression and in vivo implications. Vascular Medicine,
9, 35-45.
[14] Mandelbrot, B. B. (1983). The Fractal Geometry of Nature. New York, USA: Freeman.
[15] Kitaoka, H. and Suki, B. (1997). Branching design of the bronchial tree based on a
diameter-flow relationship. J. Appl. Phyiol., 82, 968-976.
[16] Bennett, S. H., Eldridge, M. W., Zaghi, D., Zaghi, S. E., Milstein, J. M., and Goetzman,
B. W. (2000). Form and function of fetal and neonatal pulmonary arterial bifurcations.
Am. J. Physiol. Heart Circ. Physiol., 279, H3047–H3057.
[17] Ghorishi, Z., Milstein, J. M., Poulain, F. R., Moon-Grady, A., Tacy, T., Bennett, S. H.,
Fineman, J. R. and Eldridge, M. W. (2007). Shear stress paradigm for peripheral fractal
arterial network remodeling in lambs with pulmonary hypertension and increased
pulmonary blood flow. Am. J. Physiol. Heart Circ. Physiol., 292, H3006-H3018.
 Model Methodology For Analyzing Peripheral Arterial Fractility … 237

[18] Woldenberg, M. J. (1983). Finding the optimal lengths for three branches at a junction.
J. Theor. Biol., 104, 301-318.
[19] Horsfield, K. and Woldenberg, M. J. (1989). Diameters and cross-sectional areas of
branches in the human pulmonary arterial tree. Anat. Rec., 223, 245-251.
[20] Zamir, M., Sinclair, P., and Wonnacott, T. H. (1992). Relation between diameter and
flow in major branches of the arch of the aorta. J. Biomech., 25, 1303-1310.
[21] Kassab, G. S. (2006). Scaling laws of vascular trees: of form and function. Am. J.
Physiol. Heart Circ. Physiol., 290, H894-H903.
[22] Nakamura, Y., Awa, S., Kato, H., Ito, Y. H., Kamiya, A., and Igarashi, T. (2011).
Model combining hydrodynamics and fractal theory for analysis of in vivo peripheral
pulmonary and systemic resistance of shunt cardiac defects. J. Theor. Biol., 287, 64-73.
[23] Zamir, M. (2001). Fractal dimensions and multifractility in vascular branching. J.
Theor. Biol., 212, 183-190.
[24] Grasman, J., Brascamp, J. W., Van Leeuwen, J. L., and Van Putten, B. (2003). The
Multifractal Structure of Arterial Trees. J. Theor. Biol., 220, 75-82.
[25] Mandelbrot, B. B. (1989). Multifractal Measures, Especially for Geophysicist. In: C. H.
Sholz, and B. B. Mandelbrot (Eds.), Fractals in Geophysics (pp. 5-42). Basel,
Switzerland: Birkhäuser Verlag.
[26] Michel, R. P., Hakim, T. S., Hanson, R. E., Dobell, A. R. C., Keith, F., and Drinkwater,
D. (1985). Distribution of lung vascular resistance after chronic systemic-to-pulmonary
shunts. Am. J. Physiol., 249, H1106-H1113.
[27] Mall, F. P. (1888). Die Blut und Lymphwege in Dunndarm des Hundes.
Abhandhlungen der Mathematisch-Physischen Classe der Koniglich Sachsischen
Gessellschaft der Wissenschaften, 14, 151-200.
[28] Huang, W., Yen, R. T., McLaurine, M., and Bledsoe, G. (1996). Morphometry of the
human pulmonary vasculature. J. Appl. Physiol., 81, 2123-2133.
[29] Snyder, G. K. and Sheafor, B. A. (1999). Red blood cells: centerpiece in the evolution
of vertebrate circulatory system. Amer. Zool., 39, 189-198.
[30] West, G.B., Brown, J.H., and Enquist, B.J. (1997). A general model for the origin of
allometric scaling laws in biology. Science, 276, 122-126.
[31] Fähraeus, R. and Lindqvist, T. (1931). The viscosity of the blood in narrow capillary
tubes. Am. J. Physiol., 96. 562-568.
[32] Haynes, R. H. (1960). Physical basis of the dependence of blood viscosity on tube
radius. Am. J. Physiol., 198. 1193-1200.
[33] Kamiya, A. and Takahashi, T. (2007). Quantitative assessments of morphological and
functional properties of biological trees based on their fractal nature. J. Appl. Physiol.,
102, 2315-2323.
[34] Schreiner, W., Neumann, F., Karch, R., Neumann, M., Roedler, S. M., and End, A.
(1999). Shear stress distribution in arterial tree models, generated by constrained
constructive optimization. J. Theor. Biol., 198, 27-45.
[35] Yen, R. T., Zhuang, F. Y., Fung, Y. C., Ho, H. H., Tremer, H., and Sobin, S. S. (1984).
Morphometry of cat’s pulmonary arterial tree. J. Biomech. Eng., 106, 131-136.
[36] Milnor, W. R. (1982). Steady flow. In: Hemodynamics (pp. 11-47). Baltimore, USA:
Williams and Wilkins.
238 Yoshihiro Nakamura and Shoichi Awa

[37] Hakim, T. S., Michel, R. P., and Chang, H. K. (1982). Partitioning of pulmonary
vascular resistance in dogs by arterial and venous occlusion. J. Appl. Physiol., 52, 710-
715.
[38] Rippe, B., Parker, J. C., Townsley, M. I., Mortillaro, N. A., and Taylor, A. E. (1987).
Segmental vascular resistances and compliances in dog lung. J. Appl. Physiol., 62,
1206-1215.
[39] Brody, J. S., Stemmler, E. J., and DuBois, A. B. (1968). Longitudinal distribution of
vascular resistance in the pulmonary arteries, capillaries, and veins. J. Clin. Invest., 47,
783-799.
[40] Bhattacharya, J. and Staub, N. C. (1980). Direct measurement of microvascular
pressures in the isolated perfused dog lung. Science, 210, 327-328.
[41] Meta-Greenwood, E., Meyrick, B., Steinhorn, R. H., Fineman, J. R., and Black, S. M.
(2003). Alterations in TGF-1 expression in lambs with increased pulmonary blood
flow and pulmonary hypertension. Am. J. Physiol. Lung Cell Mol. Physiol., 285, L209-
L221.
[42] Hislop, A. and Reid, L. (1973). Pulmonary arterial development during childhood:
branching pattern and structure. Thorax, 28, 129-135.
[43] Hopkins, R. A., Hammon, J. W. Jr., McHale, P. A., Smith, P. K., and Anderson, R. W.
(1979). Pulmonary vascular impedance analysis of adaptation to chronically elevated
blood flow in the awake dog. Circ. Res., 45, 267-274.
[44] Bergel, D. H. and Milnor, W. R. (1965). Pulmonary vascular impedance in the dog.
Circ. Res., 16, 401-415.
[45] Hyman, A. L. (1968). Pulmonary vasoconstriction due to nonocclusive distension of
large pulmonary arteries in the dog. Circ. Res., 23, 401-413.
[46] Rabinovitch, M. (2008). Pathophysiology of pulmonary hypertension. In: H. D. Allen,
D. J. Driscoll, R. E. Shaddy, T. F. Feltes (Eds.), Moss and Adams’ Heart Disease in
Infants, Children, and Adolescents: Including the Fetus and Young Adult (seventh
edition, pp. 1322-1354). Philadelphia, USA: Lippincott Williams and Wilkins.
[47] Hislop, A., Haworth, S. G., Shinebourne, E. A., and Reid, L. (1975). Quantitative
structural analysis of pulmonary vessels in isolated ventricular septal defect in infancy.
Br. Heart J., 37, 1014-1021.
[48] Yamaki, S., Endo, M., and Takahashi, T. (1997). Different grades of medial
hypertrophy and intimal changes in small pulmonary arteries among various types of
congenital heart disease with pulmonary hypertension. Tohoku J. Exp. Med., 182, 83-
91.
[49] Yamaki, S., Horiuchi, T., and Seino, Y. (1983). Quantitative analysis of pulmonary
vascular disease in simple cardiac anomalies with the Down syndrome. Am. J. Cardiol.,
51, 1502-1506.
[50] Heath, D. and Whitaker, W. (1957). The small pulmonary blood vessels in atrial septal
defect. Br. Heart J., 19, 327-332.
[51] Wagenvoort, C. A., Neufeld, H. N., DuShane, J. W., and Edwards, J. E. (1961). The
pulmonary arterial tree in ventricular septal defect: A quantitative study of anatomic
features in fetuses, infants, and children. Circulation 23, 740-748.
[52] Harris, P. and Heath, D. (1977). The Human Pulmonary Circulation (second edition).
Edinburgh, UK: Churchill Livingstone.
 Model Methodology For Analyzing Peripheral Arterial Fractility … 239

[53] Rossitti, S. and Lofgren, J. (1993). Vascular dimensions of the cerebral arteries follow
the principle of minimum work. Stroke, 24, 371-377.
[54] Jiang, Z. L., Kassab, G. S., and Fung, Y. C. (1994). Diameter-defined strahter system
and connectivity matrix of the pulmonary arterial tree. J. Appl. Physiol., 76, 882-892.
[55] Zhuang, F. Y., Fung, Y. C., and Yen, R. T. (1983). Analysis of blood flow in cat’s lung
with detailed anatomical and elasticity data. J. Appl. Physiol., 55, 1341-1348.
[56] Horsfield, K. (1978). Morphometry of the small pulmonary arteries in man. Circ. Res.,
42, 593-597.
[57] Singhal, S., Henderson, R., Horsfield, K., Harding, K., and Cumming, G. (1973).
Morphometry of the human pulmonary arterial tree. Circ. Res., 33, 190-197.