Documente Academic
Documente Profesional
Documente Cultură
Multidetector Computed
Tomography in Oncology
CT Perfusion Imaging
9781842143094-FM 8/8/07 12:09 PM Page ii
9781842143094-FM 8/8/07 12:09 PM Page iii
Multidetector Computed
Tomography in
Oncology
CT Perfusion Imaging
Editors
All rights reserved. No part of this publication may be reproduced, stored in a retrieval
system, or transmitted, in any form or by any means, electronic, mechanical,
photocopying, recording, or otherwise, without the prior permission of the
publisher or in accordance with the provisions of the Copyright, Designs and Patents
Act 1988 or under the terms of any licence permitting limited copying issued by the
Copyright Licensing Agency, 90 Tottenham Court Road, London W1P 0LP.
Although every effort has been made to ensure that all owners of copyright material
have been acknowledged in this publication, we would be glad to acknowledge in
subsequent reprints or editions any omissions brought to our attention.
Although every effort has been made to ensure that drug doses and other information
are presented accurately in this publication, the ultimate responsibility rests with the
prescribing physician. Neither the publishers nor the authors can be held responsible
for errors or for any consequences arising from the use of information contained herein.
For detailed prescribing information or instructions on the use of any product or
procedure discussed herein, please consult the prescribing information or instructional
material issued by the manufacturer.
A CIP record for this book is available from the British Library.
Library of Congress Cataloging-in-Publication Data
ISBN-10: 1-84214-309-3
ISBN-13: 978-1-84214-309-4
Contents
List of contributors ix
Foreword xiii
Preface xv
4 Image processing 61
Ting-Yim Lee, Xiaogang Chen, and Kenneth A Miles
viii Contents
Contributors
x Contributors
Contributors xi
Foreword
xiv Foreword
Preface
It is now more than 25 years since Leon Axel proposed a method for
determination of cerebral blood flow from rapid-sequence contrast-
enhanced computed tomography (CT). Today, the availability of rapid
imaging with multidetector CT systems and commercial analysis soft-
ware has made perfusion imaging with CT an everyday technique for
clinical practice.
CT remains an essential tool in the assessment of patients with
cancer, not only for diagnosis but also for assessment of disease extent
and severity, and for the evaluation of response to treatment. Perfusion
CT is readily performed as an adjunct to conventional CT, providing
valuable information about tumor vascularity. In many ways, perfusion
CT is not unlike CT angiography, but depicts the functional status of the
tumor circulation at tissue level rather than visualizing the morphology
of discrete vessels. By reflecting the processes of tumor angiogenesis,
this additional information can aid in diagnosis, assess tumor aggres-
sion, and help to overcome some of the limitations associated with
morphological criteria for evaluation of tumor response.
The development of perfusion CT also links with another recent
advance in cancer imaging, the introduction of integrated positron
emission tomography (PET)–CT systems. The increasing use of intra-
venous contrast material during PET–CT can be extended to include a
CT perfusion study. In this way, it is now possible to depict tumor mor-
phology, perfusion, and glucose metabolism to provide an exceptionally
detailed assessment of tumor biology in a single examination.
This book is the first to be dedicated solely to the application of
perfusion CT in oncology. The aim is to provide the technical knowledge
required to reliably obtain CT perfusion images of tumors, to give an
understanding of the pathophysiology of tumor angiogenesis and its
9781842143094-FM 8/8/07 12:09 PM Page xvi
xvi Preface
Ken Miles MD
Brighton and Sussex Medical School
9781842143094-Ch01 8/8/07 4:26 PM Page 1
1
Perfusion computed
tomography: a historical
perspective
Anne E Miles and Kenneth A Miles
(Edwin Smith Papyrus) is probably a copy of one that was first written
between 3000 and 2500 BC. Translated in 1930, this papyrus is essen-
tially a surgical document, and, for the first time in recorded history, the
word ‘brain’ is mentioned. The heart is also mentioned as the center of
a distributing system of vessels which pulsates.2 A papyrus purchased
by George Embers at Thebes in Egypt, not a copy but an original
document written about 16 centuries before the Christian Era, is now
considered to be the oldest known anatomical document. It mentions
that ‘there are vessels from it [the heart] to all the members’.2 A more
accurate impression of the structure of the human body was obtained
in ancient Egypt during the period of the New Kingdom (late dynasty
XVII through dynasty XX). This was probably possible because of the
practice of embalming and mummification which had reached its
highest level at that time.
In China, one of the oldest civilizations known, the doctrine of
Confucianism imposed restrictions upon dissection similar to those later
experienced by Galen. Dissection was not practiced, in order not to
defile the human body. Nevertheless, the medical scholars of ancient
China revealed through their writings a keen sense of awareness of the
human body for the treatment of disease. Huang Ti (2600 BC) is the
father of Chinese medicine. In his ‘Canon of Medicine’ or ‘Nei Ch’ing’
he writes that ‘all the blood of the body is under control of the heart.
The heart is in accord with the pulse. The pulse regulates all the blood
and the blood current flows in a continuous circle and never stops’.2
Remarkably, this document clearly recognizes a relationship between
blood, pulse, and the heart. It took William Harvey in the 17th century
to confirm this knowledge to the Western world.
India was yet another ancient civilization in which traditional heal-
ing methods and practical skills were remarkably advanced. Much of
this knowledge spread slowly via Asia, and reached Europe during the
Middle Ages as a result of translations that were made by Persian and
Arab scholars in the 11th century.
In the Western world, Hippocrates (about 460–377 BC), born on the
island of Cos in the Aegean, is considered by many to be the greatest
of all physicians and ‘the Father of Medicine’. However, much of the
‘Hippocratic Corpus’, a large collection of philosophical, scientific, and
medical works, was written between 300 and 200 BC by a collection
of physicians, probably of the medical school of Cos, rather than
Hippocrates himself. These writings were further compiled and edited
by the scholars of the library at Alexandria.2 Hippocrates is believed
to have disliked dissection, and his descriptions were probably based
on visual examinations of the body surface and the investigation
9781842143094-Ch01 8/8/07 4:26 PM Page 3
The ability to depict the internal anatomy of the human body through
non-invasive imaging rather than dissection or surgery dates back to the
discovery of X-rays in 1895 by Wilhelm Conrad Röntgen (1845–1923).4
Whilst Chair of Physics at the University of Würzburg, Röntgen had
been studying the phenomena associated with the passage of electric-
ity through a gas at extremely low pressure, using an evacuated glass
tube developed by Sir William Crookes (1832–1919). On November 8,
1895, Röntgen had enclosed the Crookes tube in a sealed, thick black
carton to exclude all light. When the electric current was switched on,
he noticed that a paper plate coated with barium platinocyanide began
to fluoresce, even when it was as far as 2 meters away from the tube.
He deduced the existence of hitherto unknown rays, dubbing them
‘X-rays’. The now famous radiograph of the hand of Röntgen’s wife,
Bertha, was taken within 1 month of his discovery. The shadow cast by
her bones and the ring on her finger were clearly visible, surrounded
by a penumbra produced by the flesh. In December of that year,
Röntgen published his findings in the Proceedings of the Würzburg
Physical–Medical Society in an article entitled ‘On a New Kind of Ray:
A Preliminary Communication’. By January 1896, X-rays were being
used in several countries around the world to diagnose fractures and to
detect radio-opaque foreign bodies such as bullets. The use of X-rays
for diagnosis and therapy grew rapidly thereafter. In 1901, Röntgen was
awarded the Nobel Prize for Physics in recognition of his discovery.
The means of using X-rays to depict the anatomy of the circulation
came in the 1920s with the development of cerebral angiography
by the Portuguese neurosurgeon Egas Moniz (1874–1955).5 Whilst
Professor of Neurology at Lisbon, Moniz had sought to identify a radio-
opaque dye that was non-toxic and would pass through the capillaries
without causing a blockage. Working first on animals and cadavers
before moving to human subjects, Moniz tried injections of air, bro-
mides, and iodides. The X-ray attenuating properties of the iodine atom
form the basis of contrast agents used for angiography and CT perfusion
today (Figure 1.1). However, iodine-containing compounds with lower
toxicity than the simple iodides used by Moniz were developed
9781842143094-Ch01 8/8/07 4:26 PM Page 7
Figure 1.1 Venous phase of a cerebral angiogram depicting the great cerebral
vein of Galen. Although Galen had a detailed knowledge of anatomy, he
believed that the arterial and venous systems were separate. The ability to depict
the anatomy of the cerebral circulation non-invasively came with the development
of cerebral angiography by Moniz in the 1920s
Austrian Johann Radon in 1917. At the time of his work, Hounsfield was
unaware of similar developments in the USA by the physicist Allan
McLeod Cormack. In 1979, Hounsfield and Cormack shared the Nobel
Prize for Physiology or Medicine.
Early experiments on a prototype CT scanner built by Hounsfield are
said to have included CT imaging of a cow’s head obtained from a
butcher’s shop.4 Initial images were disappointing in that they visualized
none of the details of brain structure, including the ventricles. His co-
worker, Ambrose, suggested that anatomical detail might have been
obscured by damage to the brain resulting from the blow to the head
used to kill the cow. His theory was proved correct when the experiment
was repeated using a cow’s head from a Kosher butcher, for which the
means of slaughter had been exsanguination. The images obtained on
this occasion displayed the internal brain structure with beautiful clarity.
The first clinical CT system was installed at the Atkinson Morley’s
Hospital in Wimbledon, London. The subsequent expansion of CT into
clinical practice was extraordinarily rapid. The improved visualization
of brain tumors afforded by using contrast media developed for angiog-
raphy and urography was realized very rapidly. Today, along with mag-
netic resonance imaging and ultrasound, the capacity of CT to produce
highly detailed images of the internal structure of the human body is
used not only for diagnostic purposes but also as a valuable adjunct to
the teaching of anatomy in medical education.9
The science of physiology, the concern for the internal processes and
functioning of the body as opposed to just the structure or anatomy,
really took off as a dynamic new science with William Harvey’s demon-
stration of the circulation of the blood. In fact, several people, namely
Servetus, Colombo, and Cesalpino (16th century Europeans) had
already discovered the pulmonary circulation, but it was Harvey who
became famous for publicizing it to the whole world in his book,
Exercitatio Anatomica de Motu Cordis et Sanguinis in Animalibus (1628),
which for centuries afterward was universally known as ‘de motu
cordis’. This book was written within a few years of another two
famous English books: the King James’ authorized version of the Bible
(1611) and the Folio edition of Shakespeare’s plays (1623). All three
books went down in history as essential reading in their respective
fields4. Harvey’s book was the first significant medical book ever to be
published by an English scholar.
9781842143094-Ch01 8/8/07 4:26 PM Page 9
Harvey was born in 1578 at Folkestone, England, and after his grad-
uate degree at Cambridge he travelled to Padua for his medical train-
ing, where Versalius had held the Chair of Anatomy. He returned to
London at the age of 24 and soon became the royal physician attend-
ing James I. He was described as an aloof man with a displeasing
disposition, but he was deeply respected for his scientific knowledge.
He spent his whole life dissecting as many different animals and people
as he could obtain. Harvey was very careful not to ridicule any concept
of Galen. Through his lectures and vivisections, he painstakingly
worked at repeatedly demonstrating and convincing the English medical
profession of his new ideas before he put them down in writing.
Versalius gave medicine a magnificent anatomical view of the body, but
Harvey built on Versalius’ anatomy to give medicine the full picture of
how the heart worked and how blood moved to bring life to that body.
Although Harvey gave Galen some credit for his thoughts on the circu-
lation, Harvey failed to mention that he himself had been aware of the
advanced ideas of Servetus, Colombo, and Cesalpino.
Quantitative measurements play a fundamental part in physiology,
and, interestingly, the first device for quantifying circulatory parameters
predates Harvey’s work by 25 years. In his work of 1603 entitled Method
vitandorum errorum omnium qui in arte medica contingent [Methods of
avoiding all errors pertaining to the art of medicine], Santorio Santorio
(1561–1636), later Professor of Theoretical Medicine at the University of
Padua, describes an instrument for measuring the pulse rate.10
A jump to 19th century Germany shows us the next important piece
of the jigsaw which helps to make up the complete picture of the devel-
opment of CT perfusion. Adolf Fick (1829–1901) had a remarkable
talent for mathematics and physics, but was persuaded to study medi-
cine by his elder brother, Heinrich. Heinrich, a professor of law, real-
ized that medicine would benefit from Adolf’s talents in other areas.
Soon after completing his medical degree, Fick (Figure 1.2) turned his
attentions to physiology, eventually accepting the Chair of Physiology
at Würzburg. In his Medical Physics,11 Fick introduced profound ideas
on physiological problems such as the mixing of air in the lungs, meas-
uring carbon dioxide output in humans, and the work of the heart. This
book was the first of its kind, and included studies of the hydrodynam-
ics of the circulation. Throughout his life, Fick contributed a steady
stream of information on all three disciplines of mathematics, physics,
and medicine. Even though his major work was on the physiology of
muscle contraction, he used his knowledge to demonstrate how mass
balance could be used to measure cardiac output. The concept, now
known as the Fick Principle, was published in the Proceedings
9781842143094-Ch01 8/8/07 4:26 PM Page 10
Figure 1.3 Ken Miles (left), one of the first to implement perfusion computed
tomograpy (CT) on spiral CT systems, receiving from Sir Godfrey Hounsfield
(right), the inventor of CT, a certificate commemorating the 1999 Sir Godfrey
Hounsfield Lectureship of the British Institute of Radiology
9781842143094-Ch01 8/8/07 4:26 PM Page 12
REFERENCES
1. Fishman AP, Richards DW. Circulation of the Blood: Men and Ideas. New York:
Oxford Univesity Press, 1964.
2. Persaud TVN. Early History of Human Anatomy: From Antiquity to the
Beginning of the Modern Era. Illinois: Charles C Thomas, 1984.
3. Porter R. Clinical science. In: The Greatest Benefit to Mankind: a Medical History
of Humanity from Antiquity to the Present. London: HarperCollins, 1997: 561–96.
4. Friedman M, Friedland GW. Medicine’s 10 Great Discoveries. New Haven: Yale
University Press, 1998.
5. Ferro JM. Egas Moniz (1874–1955). J Neurol 2003; 250: 376–7.
6. Doby T. Victory in an important area: cerebral angiography (1926–1931). In:
Development of Angiography and Cardiovascular Catheterization. Littleton,
MA: Publishing Sciences Group, 1976: 73–93.
7. Hounsfield GN. Computerized transverse axial scanning (tomography).
I. Description of system. Br J Radiol 1973; 46: 1016–22.
8. Ambrose J. Computerized transverse axial scanning (tomography). II. Clinical
application. Br J Radiol 1973; 46: 1023–47.
9. Miles KA. Diagnostic imaging in undergraduate medical education: an expanding
role. Clin Radiol 2005; 60: 742–5.
10. Gedeon A. Science and Technology in Medicine: an Illustrated Account Based on
Ninety-nine Landmark Publications for Five Centuries. New York: Springer, 2006.
11. Fick A. Compendium der Physiologie des Menschen mit Einschluss der
Entwickelungsgeschichte. Wien: Leipzig, 1860.
12. Moore GE. Use of radioactive diiodofluorescein in diagnosis and localization
of brain tumours. Science 1948; 107: 569–71.
13. Kety SS, Schmidt CF. The nitrous oxide method for the quantitative determi-
nation of cerebral blood flow in man: theory, procedure and normal values.
J Clin Invest 1948; 27: 476–83.
14. Penn RK, Walser R, Ackerman L. Cerebral blood volume in man. JAMA 1975;
234: 1154–5.
9781842143094-Ch01 8/8/07 4:26 PM Page 13
15. Zilkha E, Ladurner G, Iliff LD, Du Boulay GH, Marshall J. Computer subtrac-
tion in regional cerebral blood-volume measurements using the EMI-scanner.
Br J Radiol 1976; 49: 330–4.
16. Axel L. Cerebral blood flow determination by rapid-sequence computed
tomography: theoretical analysis. Radiology 1980; 137: 679–86.
17. Berninger WH, Axel L, Norman D, Napel S, Redington RW. Functional imag-
ing of the brain using computed tomography. Radiology 1981; 138: 711–16.
18. Groothius DR, Vriesendorp FJ, Kupfer B et al. Quantitative measurements of
capillary transport in human brain tumours by computed tomography. Ann
Neurol 1991; 30: 581–8.
19. Miles KA, Hayball M, Dixon AK. Colour perfusion imaging: a new application
of computed tomography. Lancet 1991; 337: 643–5.
20. Miles KA, Hayball MP, Dixon AK. Functional images of hepatic perfusion
obtained with dynamic computed tomography. Radiology 1993; 188: 405–11.
9781842143094-Ch01 8/8/07 4:26 PM Page 14
9781842143094-Ch02 8/8/07 11:54 AM Page 15
2
Scientific basis and validation
Ting-Yim Lee and Errol Stewart
INTRODUCTION
(usually 100g), and hence is the total diffusional flux across all
capillaries. It is measured in units of ml/min/100g. Expressed
in the stated units, PS can be interpreted as the unidirectional flux of
solutes from blood plasma to intersitial space that is equivalent to
the complete transfer of all the solutes in PS milliliters of blood
per minute to the interstitial space. Note that PS is implicitly tied to
the experimental set-up, in which a permeable membrane is separating
two solvents containing a solute at different concentrations. It is
equal to the diffusional flux of solute across the whole permeable
membrane (capillary endothelium) when the solvent (blood) is
‘stationary’ with respect to the membrane. For the more physiological
case of blood flowing through capillaries, the unidirectional
flux of blood-borne solutes through all the capillaries is dependent
on blood flow and PS. This relationship will be discussed in the
following.
Extraction efficiency (fraction) (E) This is the fraction of solutes
present in arterial inlets, with the potential to diffuse into the interstitial
space, that actually becomes transferred from blood to interstitial space
during a single passage of blood from the arterial end to the venous
end of the capillaries of a tumor.13 The mass of solute transferred to
the interstitial space is F ⋅ (Ca −Cv), where Ca is the arterial and Cv is the
venous concentration of the solute. The mass of solute delivered to
the tissue which can diffuse into the interstitial space is F ⋅ (Ca – Ce) until
the arterial concentration approaches the interstitial concentration, Ce.
An operational definition of E is, therefore:
C a − Cv
E= (1)
Ca − C e
Interstitial space
Ce(t), Ve
Intravascular space
FCa(t) Cb(x,t), Vb FCv(t)
PS
Figure 2.1 The Johnson and Wilson model for the distribution of blood-borne
solutes in tumor. The symbols are explained in the text. Solute concentration in
the intravascular (blood) space, Cb(x,t), is dependent on position along the cap-
illary, to reflect that it is decreasing from the arterial (Ca(t)) to the venous (Cv(t))
end of the capillary. The interstitial space is assumed to be a compartment with
no concentration gradient within it
∂C b ( x,t ) FL ∂C b ( x,t ) PS
+ + ⎡ C ( x,t ) − Ce ( t ) ⎤⎦ = 0
Vb ⎣ b
(2)
∂t Vb ∂x
For the case when Ce(t) is a constant, say Ce, equation (2) has the
solution:
⎛ V ⎞ − PS x PS
− x ⎛ V ⎞
C b (x,t) = Ca ⎜ t − b x ⎟ e FL + Ce − Ce e FL H ⎜ t − b x ⎟ (3)
⎝ FL ⎠ ⎝ FL ⎠
⎛ V ⎞ − PS ⎛ − ⎞
PS
C v ( t ) = C b ( x,t ) x = L,t ≥ Vb = Ca ⎜ t − b ⎟ e F + Ce ⎜ 1 − e F ⎟
F ⎝ F⎠ ⎝ ⎠
⎛ − ⎞
PS
Ca ( t − Tc ) − C v ( t ) = ⎜ 1 − e F ⎟ ⎡⎣ Ca ( t − Tc ) − Ce ⎤⎦
⎝ ⎠
9781842143094-Ch02 8/8/07 11:54 AM Page 19
Ca ( t − Tc ) − C v ( t ) −
PS
E≈ = 1− e F (4)
Ca ( t − Tc ) − Ce
dCe ( t )
= F ⎡⎣Ca ( t − Tc ) − C v ( t ) ⎤⎦
dt
dCe ( t )
Ve = FE ⎡⎣Ca ( t − Tc ) − Ce ⎤⎦
dt
The last equation can be interpreted as that the forward flux from the
capillary to the interstitial space is FE·Ca(t−Tc), and the backflux from
the interstial space to the capillary is FE⋅Ce. Thus, FE is the unidirec-
tional flux of solute per unit concentration, or transfer constant, from
blood to interstitial space or from interstitial space to blood. The above
derivation is obtained under the special case when Ce(t) is held con-
stant in time. For the general case when Ce(t) is an arbitrary function of
time, St Lawrence and Lee7 have shown that the unidirectional flux of
solute per unit concentration is still FE.
There exist three regimes for the exchange of solute between
blood and interstitial space: (1) when PS <<F, FE approximates PS,
the exchange is diffusion limited; (2) when PS >>F, so that FE
approaches F, the exchange is flow limited; and, (3) when PS is of
the same magnitude as F, the exchange is neither diffusion nor flow
limited.
Vb
F=
Tm
9781842143094-Ch02 8/8/07 11:54 AM Page 20
In tracer kinetics modeling, the impulse residue function (IRF)24 and the
mathematical operation of convolution are frequently used to simplify
and provide insight into the dual processes of the convective transport
by blood flow of intravenously administered iodinated contrast agent to
the tumor, and the exchange by diffusion of these contrast molecules
between the intravascular space and the extravascular interstitial space.
In the following, the IRF will be defined and the convolution operation
explained in more detail.
9781842143094-Ch02 8/8/07 11:54 AM Page 21
a b
1.2 70
Impulse residue function (no units)
1 60
50
0.8 Vb
40
0.6
30
0.4 Vb
Vb 20
Tmin = Tm = Tmin = Tm =
F F
0.2 10
0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
Time (s) Time (s)
Figure 2.2 (a) The impulse residue function according to the Johnson and
Wilson model. The symbols are defined in the text. (b) The blood flow scaled
impulse residue function is obtained by scaling the impulse residue function in
(a) by blood flow, which in this example is 60 ml/min/100 g
9781842143094-Ch02 8/8/07 11:54 AM Page 22
mean transit time (Tm). As the mass of contrast agent injected is unity,
the plateau height in the vascular phase is also unity. The drop at Tmin
represents the fraction of contrast injected that is not extracted by the
tumor and is carried out with the venous blood. The second or intersti-
tial phase begins at Tmin and shows a slow decay towards baseline. The
initial height of the second or interstitial phase is the fraction of
contrast that is extracted by the tumor, or it is equal to the extraction
efficiency of contrast by the tumor. The slowly decaying interstitial
phase represents the return of the extracted contrast from the tumor
to the bloodstream and then clearance via blood flow. The tumor
IRF can be interpreted as the fraction of contrast that remains (‘resides’)
in the tumor after an ‘impulse’ injection of unit mass of contrast, and
as such is unitless. If the amount deposited is M0 instead, then the
corresponding tissue residue function is the product of M0 and the
IRF and in this case M0·IRF has the same units as the tissue
residue function. The IRF is a theoretical concept and cannot be meas-
ured easily in clinical practice since it requires, as a close approxima-
tion, an intra-arterial bolus injection into one of the supply arteries of
the tumor.
Convolution
An intravenous injection of contrast gives rise to an arterial input TDC,
Ca(t), which is not a delta function, and the corresponding tumor
residue function or tumor time–density curve, Q(t), is related to Ca(t)
via the impulse residue function as follows:17
a b
1.2 2.0
Arterial conc. (HU/ml)
c d
200 10
Arterial conc. (HU/ml)
120 6
80 4
40 2
0 0
0 15 30 45 60 75 90 105 0 15 30 45 60 75 90 105
Time (s) Time (s)
e f
350 30
Arterial conc. (HU/ml)
300 25
250 20
200
15
150
10
100
50 5
0 0
0 15 30 45 60 75 90 105 0 15 30 45 60 75 90 105
Time (s) Time (s)
at each of the injection times. The resultant tumor TDC (Figure 2.3f)
in response to the general arterial concentration Ca(t) is the sum of
all the scaled IRFs after they have been shifted in time in accordance
with the times of their corresponding bolus injections. In summary, if
the IRF is known, the corresponding tumor TDC in response to
the arterial TDC, Ca(t), can be obtained as a summation of scaled and
time shifted IRFs. The scale factors and time shifts are given by
F⋅Ca(t) and t. This operation, as illustrated in Figure 2.3f, is called a
convolution.
Q( t ) = Ca ( t ) ⊗ ⎡⎣ F ⋅ R ( t ) ⎤⎦ (5a)
where F·R(t) is the blood flow scaled impulse residue function (FIRF).
Note that whereas IRF is unitless, FIRF has the unit of blood flow:
ml/min/100g. Convolution makes it possible in our discussion to
separate the influence of the details of the administration of contrast
from the effects of the inherent properties of the tumor on the tumor
residue function, Q(t). The arterial input TDC in equation (5a), Ca(t), is
governed by factors such as injection rate and dosage related to the
contrast injection. It is also affected by factors related to the central
hemodynamics of the subject, such as ejection fraction and heart
rate. The blood flow scaled impulse residue function, F⋅ R(t), reflects
tumor blood flow F and other inherent properties of the tumor.
Furthermore, for inert contrast agent, it is possible as shown in the
following section to determine the functional (mathematical) form of
the FIRF in terms of tumor blood flow (F), blood volume (Vb), mean
transit time (Tm), distribution volume (Ve), and capillary permeability
surface area product (PS) or extraction efficiency (E) of contrast agent.
In CT tumor perfusion studies, both Q(t) and Ca(t) are measured, and
the FIRF is convolved with the measured Ca(t) to obtain a predicted
(fitted) tumor residue function, Q̃(t). Model deconvolution of the flow
scaled impulse residue function from the measured Q(t) and Ca(t) deter-
mines values for the set of parameters: F, Vb, Ve, and PS (E) iteratively
by adjusting their values, and hence the corresponding FIRF, and
checking the deviations of Q̃(t) from Q(t) at each iteration until the
minimum is achieved.
9781842143094-Ch02 8/8/07 11:54 AM Page 25
Fick principle
Considering a mass of tumor, F is the perfusion, Ca(t) is the contrast
concentration in the arterial inlet(s), and Cv(t) is the contrast concentra-
tion in the venous outlet(s). At time T, the accumulated mass of contrast
from the arterial in-flow, Qin(T), is:
Qin ( T ) = F ⋅ ∫ Ca ( t ) ⋅ dt
0
Q out ( T ) = F ⋅ ∫ C v ( t ) ⋅ dt
0
Q( T ) = F ⋅ ∫ ⎡⎣ Ca ( t ) − C v ( t ) ⎤⎦ ⋅ dt (6)
0
Equation (6) states that the accumulated mass of contrast in the tumor
over a time period [0, T] is equal to the product of tumor perfusion and
the time integral of the arteriovenous difference in contrast concentration.
No outflow assumption
One immediate simplification is to assume that during the time period
[0, T] the venous concentration, Cv(t), or the venous outflow, is
9781842143094-Ch02 8/8/07 11:54 AM Page 26
Q( T ) = F ⋅ ∫ Ca ( t ) ⋅ dt
0
⎡ dQ( t ) ⎤
⎢ dt ⎥ = F ⋅ Ca ( T ) (7)
⎣ ⎦t=T
⎡ dQ( t ) ⎤
⎢ dt ⎥ = F ⋅ ⎡⎣ Ca ( t ) ⎤⎦ max (7a)
⎣ ⎦ max
Equation (7a) states that tumor perfusion is the ratio of the maximal rate
of accumulation of contrast in tissue or the maximum slope of Q(t) to
the maximum arterial concentration. For this reason, calculation based
on equation (7a) is also called the maximum slope method. Also, in
order to satisfy the no venous outflow assumption, a relatively high
injection rate (15–20 ml/s) has to be used.28
To avoid the no venous outflow assumption, Cv(t) has to be meas-
ured. This will be possible only if we can identify draining veins. If we
are interested in measuring blood flow of the whole organ, for exam-
ple the brain, then the jugular vein can be used. This technique was
used by Kety and Schmidt to measure global brain blood flow in human
subjects in the 1940s.29 For regional tumor blood flow measurement,
the local draining veins have to be identified and contrast concentration
in them measured. In general, this is not possible. When there is signif-
icant outflow of contrast with venous blood, equation (6) suggests that
the true perfusion would be underestimated by equation (7a).
KINETICS MODELING
Compartment models
A good example of the use of compartmental models to describe the
distribution of blood-borne solutes in tissue was the study by Blasberg
and Patlak.30,31 Groothuis et al.32,33 were the first to apply compartmen-
tal models to describe the distribution of X-ray contrast agent in tumor,
and used CT to study this distribution in canine and human brain
tumors. Since X-ray contrast agents are hydrophilic, they are excluded
from the intracellular space both in blood and in parenchymal tissue.
Furthermore, because contrast agents are usually inert (i.e. not metab-
olized) in tissue, the modeling of the distribution of contrast agent
in tissue calls for only two compartments, blood (intravascular) and
interstitial space (Figure 2.4). The following equation can be written, by
applying the Fick principle to the interstitial space:
dCe ( t )
Ve = K1 ⋅ C b ( t ) − k 2 ⋅ C e ( t ) (8)
dt
where Cb(t) and Ce(t) are the blood and interstitial concentrations
of contrast agent (solute) respectively. Ve is the distribution volume
of contrast agent in the interstitial space. K1 is the forward transfer
constant from the intravascular space into the interstitial space
and k2 is the backflux constant from interstitial space to intrasvascular
space. As discussed previously, for contrast agent, both the forward and
Capillary
endothelium
K1
∫ C ( u) ⋅ e
− k⋅ ( t−u )
Ce ( t ) = k ⋅ a
du (9)
0
∫ C ( u) ⋅ e
− k⋅ ( t−u )
Ce ( t ) = k ⋅ a
du + Ca ( t ) ⋅ Vb (10)
0
Q( t ) = Ca ( t ) ⊗ [ FE ⋅ e-kt +V b ⋅ δ( t )] (11)
where δ(t) is the delta function. By comparing equations (5a) and (11),
the FIRF for the two-compartment model shown in Figure 2.4 is:
Equation (12) for the blood flow scaled impulse residue function then
simplifies to:
F ⋅ R ( t ) = FE ⋅ H( t ) + V b ⋅δ( t ) (12a)
where H(t) is the unit step function and δ(t) is the delta function. The
tumor residue function, Q(t), simplifies to:
t
Q( t ) = FE ∫ Ca ( u )du + Vb⋅ Ca ( t )
0
Q( t ) ∫0 Ca ( u )du (11a)
= FE ⋅ + Vb
Ca ( t ) Ca ( t )
∂C b ( x,t ) FL ∂C b ( x,t ) PS
+ + ⎡ C ( x,t ) − Ce ( t ) ⎤⎦ = 0
Vb ⎣ b
(13a)
∂t Vb ∂x
dCe ( t ) PS L
L ∫0 ⎣ b
Ve = ⎡C ( x,t ) − Ce ( t ) ⎤⎦ dx (13b)
dt
While the measurement of the mass of contrast agent per unit mass of
tissue, Q(t), is relatively simple because of the inherent linear relation-
ship between CT enhancement and contrast agent concentration,
there are a number of additional factors requiring attention with
respect to the determination of Ca(t). These factors include: (1) partial
volume averaging; (2) possible dispersion of the arterial concentration
function between the site of measurement and the site of input into
the tissue;36 (3) delay in arrival at the input relative to the measure-
ment site;36 and (4) the fact that some tissues can have more than one
input; for example, the liver, besides the hepatic artery input, also
receives blood from the portal vein, which drains venous blood from
the gut.
m
where k is less than or equal to one. When Ca ( t ) instead of Ca(t) is
used in equation (5a), the measured blood flow, Fm, is equal to the true
blood flow divided by kpva (F/kpva). Since kpva is less than one, Fm is
greater than the true F when PVA affects the measurement of Ca(t).
In brain tumor studies, since all of the intracranial arteries that can be
used for Ca(t) are small, PVA is expected to be significant, or kpva is less
than one in most cases. To correct for the effect of PVA is equivalent
to determining a value of kpva. If there exist regions in veins, for
example the superior sagittal sinus, which are free of PVA, then unlike
arterial curves, venous curves, Cv(t), can be measured accurately. Let
h(t) be the transit time spectrum from artery to vein,36 then
1
C v ( t ) = Ca ( t ) ⊗ h( t ) = ⋅ ⎡ Cm ( t ) ⊗ h( t ) ⎤⎦
k pva ⎣ a
∞ ∞
1
A v = ∫ C v ( u )du = ⋅ ∫ ⎡⎣C
m
a
( u ) ⊗h( u )⎤⎦ du (16)
0
k pva 0
∫ h( u )du = 1
0
and
∞ ∞
∫ ⎡⎣ Ca ( u ) ⊗ h( u ) ⎤⎦ ⋅ du = ∫ Ca ( u ) ⋅ du = A a
m m m
0 0 (17)
m
where A a is the area underneath the measured arterial curve. From
equations (16) and (17), k is determined as the ratio:
A am
k pva =
Av
9781842143094-Ch02 8/8/07 11:54 AM Page 33
Ca ( t ) = Cam ( t ) ⊗ h( t ) (18)
Substituting equation (18) into equation (5a), the operating equation for
model deconvolution:
arterial concentration function. To account for the finite transit time, T0,
between the measurement region and the input to the tumor:
Ca ( t ) = Cam ( t − T0 )
F ⋅ R ( t − T0 ) = FE ⋅ e ( 0 ) + Vb ⋅ δ ( t − T0 )
− k t−T
(21)
⎧0 0 ≤ t ≤ T0
⎪
F ⋅ R ( t − T0 ) = ⎨F T0 ≤ t ≤ T0 + Tm (23)
⎪ -k( t-T0-Tm )
⎩FE ⋅ e ⋅ H( t − T0 − Tm ) t > T0 + Tm
blood flow to the liver is supplied by the portal vein, and the remain-
ing one-third is supplied by the common and proper hepatic arteries.38
The hepatic artery input, Ch(t), can be approximated by the aortic input
at the level of the liver, while the portal vein input, Cp(t), can be directly
measured if one of the CT slices includes the portal vein in its field of
view. If α is the fraction of total blood flow to liver tissue that arises
from the hepatic artery (or hepatic arterial fraction), then Ca(t) can be
expressed as the weighted sum of Ch(t) and Cp(t):39
Ca ( t ) = α ⋅ Ch ( t ) + (1 − α ) ⋅ Cp ( t ) (24)
By replacing Ca(t) in equations (5a), (21), and (23) with that expressed
in equation (24), the hepatic arterial fraction can be estimated together
with other parameters in the two-compartment and Johnson and Wilson
models.
Recirculation
The effect of recirculation on the kinetics modeling proposed above
can be analyzed in a straightforward manner as follows. To account
for recirculation, the arterial TDC, Ca(t), can be written as a summation
of the first pass component, C FP a (t), and subsequent recirculation
components. For simplicity we will assume that only the first recircula-
tion component, C aR1(t), is important, and the following derivation
can easily be generalized to more than one recirculation component.
If tumor perfusion is unchanged (stationary) for the duration of the
first pass and subsequent recirculation phases, then equation (5a)
would apply to all phases with the same blood flow scaled impulse
residue function, F·R(t). The tissue residue function for the first
pass phase, QFP(t), and the first recirculation phase, QR1(t), can be
written as:
QFP (t) + QR1 ( t ) = ⎡⎣Ca FP (t) + CaR1 ( t )⎤⎦ ⊗ ⎡⎣F ⋅ R(t) ⎤⎦ (26)
9781842143094-Ch02 8/8/07 11:54 AM Page 36
Since, by definition:
Q(t) = Q FP (t) + Q R1 ( t )
which is the same as equation (5a). Thus, provided that blood flow is
stationary, the blood flow scaled IRF, F·R(t), can be calculated by model
deconvolution between the measured arterial and tissue TDCs, which
contain both the first pass and subsequent recirculation phases. There
is no need to correct for recirculation in the measured TDCs.
PRACTICAL ISSUES
For the Johnson and Wilson model, the blood flow scaled impulse
residue function has two distinct phases. The first or vascular phase,
as shown in Figure 2.2, is relatively brief, being equal to the mean tran-
sit time through the tumor vasculature, which is in seconds rather than
minutes. The second or interstitial phase is a slow exponential decay
with a half-life which is usually measured in minutes. A two-phase
scanning protocol is required for the Johnson and Wilson model. The
first phase of scanning has an image interval of 0.5–1 s and a duration
of 30–45 s to allow reliable determination of the vascular phase of
the FIRF. The short image interval in the first phase requires the X-ray
to be on all the time, to scan continuously, and as such constitutes
the bulk of the radiation dose to the patient. The brevity of the
duration of the first phase means that breath-holding required for
studies affected by breathing motion during this phase of scanning is
possible. The second phase of scanning has an image interval of
10–20 s and duration of several minutes to characterize the rate
constant, k, in equation (23). At this infrequent image interval, X-rays are
turned on intermittently to acquire a scan. Stewart et al. have proposed
an intermittent scanning method for this second phase that allows the
correction of breathing motion without respiratory gating. For reasons
of space, interested readers are referred to their original publication.39
The following is a typical two-phase scanning protocol, as an
example:
• first phase: image interval 1 s and 30 s duration
• a delay of 15 s between first and second phases
• second phase: image interval 15 s and 2 min duration.
At each CT slice location, the first phase acquires 30 images
while the second phase acquires only eight images. Thus, the first
phase contributes close to 80% of the total radiation dose if the
same X-ray tube current (mA) is used in both phases. We can also
compare the radiation dose required by the two-compartment and
Johnson and Wilson models. For the same duration of 2.5 min and at
15s image interval, the two-compartment model protocol acquires
10 images per slice location while the Johnson and Wilson protocol
acquires 38. The radiation dose for the Johnson and Wilson
model scanning protocol is 3.8 times higher than that of the two-
compartment model, again assuming that the same X-ray tube current
is used to acquire each image. On the other hand, the advantage of the
Johnson and Wilson model is that absolute tumor perfusion can be
measured.
9781842143094-Ch02 8/8/07 11:54 AM Page 38
Calibration of CT scanner
In CT tumor perfusion studies, there is an implicit assumption that
enhancement measured by a CT scanner in HU in artery and tumor is
linearly proportional to their contrast concentration, Ca(t) and Q(t) in
equation (5a). As long as the proportional relationship is the same for
both artery and tumor, calibration of enhancement in Hounsfield units
(HU) versus contrast concentration is not required. This is true for
CT scanners when beam hardening is negligible. For brain tumor per-
fusion studies, given the relatively low enhancement in arteries and
veins in the brain at the dosage of contrast injected,41 this condition is
usually satisfied. For other tissues, particularly in the abdominal or tho-
racic region, the dosage of contrast injected may have to be limited to
reduce the beam hardening effect arising from the superior vena cava
or the aorta, if the injection site is any one of the upper-extremity veins.
The accuracy and precision of CT tumor blood flow and related meas-
urements has been investigated in animal models of brain tumor, soft
tissue tumor, and liver tumor, using methods that do not require mod-
eling, as well as kinetics modeling methods. These preclinical studies
have the advantage that the accuracy of CT measurements can be
validated against reference gold-standard microsphere measurements,
which require tissue sampling and hence sacrifice of the animal. In this
section, the results from these validation studies are discussed.
blood volume, and mean transit time was 14%, 18%, 20% and 24%,
respectively.
T0, contrast arrival time; HBF, hepatic blood flow; HBV, hepatic blood volume; PS, permeability surface
area product; HAF, hepatic arterial fraction
Reproducibility studies
Goh et al.52 investigated the reproducibility of measurements of blood
flow, blood volume, mean transit time, and permeability surface area
product obtained with the Johnson–Wilson model (equations (5a)
and (23)) in 10 patients with colorectal cancer. Each patient underwent
two single-phase 65-s perfusion studies separated by 48 hours to allow
assessment of reproducibility of the CT measurements listed above. The
mean difference (95% limits of agreement) for blood volume, blood flow,
mean transit time, and permeability when a 5-mm slice thickness was used
were 0.04 (−2.50 to +2.43)ml/100g +8.80 (−50.5 to +68.0)ml/min/100 g;
–0.99 (−8.19 to +6.20)s; and +1.20 (−5.42 to +7.83)ml/min/100 g,
respectively. Similar reproducibility results were obtained for a 20-mm
slice thickness.
Ng et al.53 investigated the reproducibility of measurements of blood
volume and permeability surface area product obtained using a two-
compartment model (Figure 2.4) under the assumption that there is no
backflux of contrast agent from the interstitial to the intravascular space,
that is, equations (5a) and (22). As discussed before, the FE estimated
from equations (5a) and (22) can be equated to the PS of the tumor
only when PS is much smaller than blood flow (F). In their study, blood
flow was not estimated, and thus it is not certain whether the estimated
FE is equal to PS. Nevertheless, the mean difference (95% limits of
agreement) for tumor FE was 1.4 (−4.0 to 6.8)ml/min/100 ml for 10-mm
slice thickness and 0.8 (−3.6 to 5.2)ml/min/100 ml for 40-mm slice
thickness. The mean difference (95% limits of agreement) for blood
volume was 1.9 (−5.1 to 8.9)ml/100ml for 10-mm slice thickness and
9781842143094-Ch02 8/8/07 11:54 AM Page 43
Accuracy studies
Hattori et al.54 compared CT tumor perfusion measurement obtained by
application of the Fick principle with the no outflow assumption
(equation (7a)) against that measured with positron emission tomogra-
phy using the [15O-]carbon dioxide (C15O2) steady-state method in 16
patients with superficial tumors including the sites of larynx, hypo-
pharynx, tongue, lung, and breast. The CT blood flow measurement on
average was about 20% higher than the PET measurement, and the cor-
relation coefficient between the two sets of measurements was high
(r=0.79). These results suggest that CT tumor blood flow measurement
in patients using the no venous outflow assumption is accurate, albeit
with a systemic overestimation of about 20%.
CONCLUSION
REFERENCES
39. Stewart EE, Chen X, Hadway J, Lee T-Y. Correlation between hepatic tumor
blood flow and glucose utilization in a rabbit liver tumor model. Radiology
2006; 239: 740–50.
40. Roberts H, Roberts T, Lee T-Y, Dillon W. Dynamic, contrast-enhanced CT of
human brain tumors: quantitative assessment of blood volume, blood flow,
and microvascular permeability. Report of Two Cases. AJNR Am J Neuroradiol
2002; 23: 828–32.
41. Nabavi DG, LeBlanc LM, Baxter B et al. Monitoring cerebral perfusion flow
after subarachnoid hemorrhage using CT. Neuroradiology 2001; 43: 7–16.
42. Lee TY, Ellis RJ, Dunscombe PB et al. Quantitative computed tomography of
the brain with xenon enhancement: a phantom study with the GE9800
scanner. Phys Med Biol 1990; 35: 925–35.
43. Wintermark M, Maeder P, Verdun FR et al. Using 80 kVp versus 120 kVp in
perfusion CT measurement of regional cerebral blood flow. AJNR Am
J Neuroradiol 2000; 21: 1881–4.
44. Hirata M, Sugawara Y, Fukutomi Y et al. Measurement of radiation dose in
cerebral CT perfusion study. Radiat Med 2005; 23: 97–103.
45. Huda W, Scalzetti EM, Levin G. Technique factors and image quality as functions
of patient weight at abdominal CT. Radiology 2000; 217: 430–5.
46. Pollard RE, Garcia TC, Stieger SM et al. Quantitative evaluation of perfusion
and permeability of peripheral tumours using contrast-enhanced computed
tomography. Invest Radiol 2004; 39: 340–9.
47. Heymann MA, Payne BD, Hoffman JI, and Rudolph AM. Blood flow measure-
ments with radionuclide-labeled particles. Prog Cardiovasc Dis 1977; 20: 55–79.
48. Lykke AW, Cummings R. Increased vascular permeability in the primary
cutaneous allograft response in the rat. Experientia 1969; 25: 1287–8.
49. Purdie TG, Henderson E, Lee TY. Functional CT imaging of angiogenesis in
rabbit VX2 soft tissue. Phys Med Biol 2001; 46: 3161–75.
50. Bland JM, Altman DG. Statistical methods for assessing agreement between
two methods of clinical measurement. Lancet 1986; 1: 307–10.
51. Eliasziw M, Young SL, Woodbury MG, Fryday-Field K. Statistical methodology
for the concurrent assessment of interrater and intrarater reliability: using
goniometric measurements as an example. Phys Ther 1994; 74: 777–88.
52. Goh V, Halligan S, Gartner L, Bassett P, Bartram CI. Quantitative colorectal
cancer perfusion measurement by multidetector-row CT: does greater tumour
coverage improve measurement reproducibility? Br J Radiol 2006; 79: 578–83.
53. Ng QS, Goh V, Klotz E et al. Quantitative assessment of lung cancer perfusion
using MDCT: does measurement reproducibility improve with greater tumor
volume coverage? Am J Roentgenol 2006; 187: 1079–84.
54. Hattori H, Miyoshi T, Okada J et al. Tumor blood flow measured using
dynamic computed tomography. Invest Radiol 1994; 29: 873–6.
9781842143094-Ch03 8/8/07 11:54 AM Page 47
3
Image acquisition and contrast
enhancement protocols for
CT perfusion
Kenneth A Miles
CT perfusion
Time
Image acquisition
Overall time 60 s 50 s 90 s
Number of images 60 ¥4 25 ¥2 9¥6
Image frequency Every 1 s Every 2 s Every 10s
No. slices ¥thickness Single location: Single location: Multiple spiral:
4 ¥5 mm 2 ¥10 mm 6¥10 mm
Tube voltage 120 kVp 80 kVp 80 kVp
Tube current 50 -100 mAs 100 -200 mAs 100 –200 mAs
Contrast medium
Concentration 370 mg/ml 370 mg/ml 300 mg/ml
Volume 50 ml 40 ml 100 ml
Injection rate 4–7 ml/s 7–10 ml/s Decreasing rate:
4ml/s–2ml/s–1ml/s
Processing
Parameters calculated Perfusion, Perfusion, Standardized
blood volume, blood volume, perfusion value,
mean transit mean transit time permeability,
time, permeability blood volume
Analysis method Deconvolution Single Two compartment
compartment (Patlak)
9781842143094-Ch03 8/8/07 11:54 AM Page 50
Frequently, the desire for image data of higher quality must be bal-
anced against this radiation dose. Although, in the context of oncology,
the radiation exposure associated with CT perfusion is small compared
to the radiotherapy dose that many patients will receive, there remains
a need to limit the radiation burden associated with CT perfusion
studies. A typical radiation dose received from a CT perfusion study of
a limited volume (e.g. 4¥5 mm slices) is between 2 and 10mSv, depend-
ing on the body region, and is therefore comparable to that obtained
from a gamma camera study of tumor perfusion using single photon
emission tomography, for instance. As multidetector CT systems are
developed with larger detector tracks, offering the possibility of single-
location acquisitions over larger tumor volumes, the radiation dose
associated with CT perfusion could potentially increase by a
considerable amount. It should also be remembered that the aim of
CT perfusion is to produce physiological data rather than display
anatomical structure, and, thus, spatial resolution, although important,
is not the primary consideration.
IMAGE ACQUISITION
Phase of respiration
Respiratory motion can result in significant misregistration of images
acquired during the dynamic image sequence for CT perfusion in the
chest and abdomen. Misregistration leads in turn to errors in perfusion
values and artifacts on parametric images. CT perfusion protocols for
chest and abdomen are usually acquired with either suspended respi-
ration or quiet breathing. Suspended respiration is unlikely to be appro-
priate for image series longer than 45–60 seconds, unless acquisition
pauses are created to allow the patient to take a breath, as might be the
case for multiple spiral protocols. Many patients who have suspended
their breathing will slowly exhale during the acquisition, resulting in
motion artifacts. Protocols that adopt quiet respiration can result in
high-quality perfusion images, but the patient must be warned to avoid
the temptation to take a deep breath when experiencing the ‘hot flush’
commonly associated with a rapid bolus of contrast medium. Quiet res-
piration will also be appropriate when co-registering CT perfusion data
with PET images, acquired using an integrated PET–CT system, as
suspended respiration is not possible for the PET images, which take
several minutes. Respiratory gating of CT images has the potential
to reduce CT perfusion motion artifacts in the future.
9781842143094-Ch03 8/8/07 11:54 AM Page 51
parameters reduce image noise but increase radiation dose for the
patient. Typical exposure times for current CT scanners are 1 second or
less. Selecting a smoother reconstruction filter (e.g. soft-tissue) will
reduce image noise, but will also reduce spatial resolution.
The tube voltage also governs the amount of attenuation produced
by a given iodine concentration of contrast medium (Figure 3.3).
A greater increase in attenuation is produced at tube voltages lower
than that typically used for diagnostic studies (e.g. 80–100kVp instead
of 120–140 kVp). However, a lower tube voltage heightens image noise
and beam hardening effects. The precise relationship between attenua-
tion change and iodine concentration varies between CT systems, with
aging of the X-ray tube, and for different slice positions across the track
of a multidetector CT system.13 The likely explanation for these varia-
tions in iodine response is differences in the X-ray spectra generated by
a particular X-ray tube. The X-ray spectrum will be modified by beam
hardening within the anode, which will not only change as the anode
wears over time but also vary across the X-ray beam (anode heal
effect). Because algorithms for calculating perfusion from CT typically
divide tissue enhancement measures by vascular enhancement, the
effects of variable iodine response will cancel out for a given perfusion
measurement, provided that the tissue and vascular enhancement data
are obtained from the same portion of the detector track.
The different analysis methods used in processing CT perfusion data
(see Chapter 2) are more or less sensitive to image noise, and thus
the X-ray exposure should be matched to the processing algorithm.
800
600
Attenuation (HU)
400
80 kVp
200
100 kVp
120 kVp
140 kVp
0
0 5 10 15 20
Iodine concentration (mg/ml)
CONTRAST ENHANCEMENT
2. Maximized tissue
HU
contrast enhancement
Tissue
Time
CHOICE OF PROTOCOL
Tumor enhancement
Tumor enhancement values are more readily compared when expressed
as a ratio of enhancement to injected dose of contrast material (e.g. HU/g
iodine). However, it is important to compare other factors such as rate
of injection, time of image acquisition, and, most important, the tube
voltage used. A reduction in tube voltage from 120kVp to 80kVp can
increase the sensitivity of a CT system to iodine by as much as 50%.13
For example, let us compare the technique for assessing the likelihood
of malignancy within pulmonary nodules proposed by Swensen et al.7
to the study by Shim et al. who used enhancement of T1 lung cancers to
predict the likelihood of nodal metastases.17 Swensen et al. adopted an
enhancement threshold of 15 HU, whereas Shim et al. used 60 HU. Both
studies used a tube voltage of 120kVp. In the Swensen study, a typical
70-kg man would have received 29.4 g of iodine injected over 49 s, and
thus the enhancement threshold could be expressed as 0.51HU/g iodine.
Shim et al. used a 60-HU enhancement threshold following an injection of
36g of iodine administered over a similar time (40 s), giving a corrected
threshold of 1.67HU/g iodine. It can be seen that, although the enhancement
threshold associated with an increased likelihood of nodal metastases is
four times that needed to diagnose malignancy, this difference is nearer
three times when enhancement is corrected for the dose of iodine
injected.
SUMMARY
REFERENCES
4
Image processing
Ting-Yim Lee, Xiaogang Chen, and Kenneth A Miles
INTRODUCTION
The methods for deriving tumor perfusion data from dynamic contrast-
enhanced computed tomography (CT) images outlined in the previous
chapters can be applied to regions of interest (ROI) constructed on the
basis of anatomical features visible on conventional CT images.
However, abnormalities of tumor perfusion may not be readily appar-
ent on conventional CT images, and therefore the ability to place ROIs
over areas of physiological importance is severely limited. If perfusion
values are calculated using time–attenuation data derived from individ-
ual pixels, it is possible to generate a parametric map depicting blood
flow throughout the whole CT slice chosen for the dynamic study.
The spatial resolution of these CT perfusion images is identical to that
of conventional CT, and thus superior to other techniques such as
positron emission tomography or even magnetic resonance imaging.
Indeed, in addition to images of tumor blood flow, a whole range of
parametric maps can be produced in this way (Table 4.1). Figure 4.1
shows the appearance time (Figure 4.1a), blood flow (Figure 4.1b),
blood volume (Figure 4.1c), and capillary permeability surface area
product (Figure 4.1d) of a brain tumor in the right cerebral hemisphere.
The tumor (arrow in Figure 4.1d) is clearly delineated in the blood flow
(F), blood volume (Vb), and capillary permeability surface area product
(PS) parametric maps as having a hypervascular rim surrounding a
hypovascular core, with high blood flow, blood volume, and capillary
permeability surface area product in the rim but low values in the core.
There are also two distant metastases (arrowheads in Figure 4.1d)
which are visible as focal regions of increased PS and blood flow. In
the normal brain, there is clear differentiation between gray and white
matter in both blood flow and blood volume. In the appearance time
map, the distinction between gray and white matter is less, suggesting
9781842143094-Ch04 8/8/07 11:55 AM Page 62
Operational equations
Parameter Method of derivation for derviation*
*Refer to Chapter 2 for discussion of the equations listed. For liver studies, equation (24) has to be
included.
†
As discussed in Chapter 2, for F >> PS, FE ≈ PS; and for F << PS, FE ≈ F.
that the arrival times for gray and white matter are more similar than
their blood flows and volumes.
In some studies, the visualization of areas of physiological difference
can be obscured by image noise. Examples include the higher perfusion
values within cerebral cortex as compared to white matter (Figure 4.2a),
and tumors with perfusion values close to that of the surrounding tissue
(Figure 4.3a). Performing a spatial smoothing process can reduce the
impact of noise by averaging out the individual pixel values over a
larger area, and so displays physiological differences more clearly
(Figures 4.2b and 4.3b). It is usual for a weighted average to be
adopted, such that the pixel values immediately adjacent to the pixel
under consideration are weighted more heavily than pixels further
away. An example is 3 × 3 binomial smoothing whose weighting
factors are summarized in Table 4.2. In this example, the central pixel
9781842143094-Ch04 8/8/07 11:55 AM Page 63
Image processing 63
a b
5 90
0 0
c d
5 30
0 0
Figure 4.1 Functional maps from a computed tomography (CT) tumor perfu-
sion study in a patient with a glioblastoma multiforme in the right cerebral
hemisphere: (a) appearance time (T0) map, pixel values in units of seconds;
(b) blood flow (F) map, pixel values in units of ml/min/100 g; (c) blood volume
(Vb) map, pixel values in units of ml/100 g; (d) capillary permeability surface
area product (PS) map, pixel values in units of ml/min/100 g. The arrow points
to the main tumor whilst the arrowheads indicate two distant metastases
a b
90 90
0 0
Figure 4.2 Blood flow maps of the same patient as in Figure 4.1 without
(a) and with (b) binomial spatial smoothing. Note that the higher perfusion
within gray matter structures is more readily apparent on the smoothed image
9781842143094-Ch04 8/8/07 11:55 AM Page 64
a b
Figure 4.3 Hepatic arterial perfusion maps displayed without (a) and with (b)
binomial spatial smoothing. Note that the higher perfusion within the hepatic
metastasis (arrow) is more readily apparent on the smoothed image
1 N
F= ∑F
N i=1 i
(1a)
1 N
V= ∑V
N i=1 b,i
(1b)
Table 4.2 Weighting factors for individual pixels for 3 × 3 binomial spatial smoothing
1 2 1
2 4 2
1 2 1
9781842143094-Ch04 8/8/07 11:55 AM Page 65
Image processing 65
where Vb,i is the blood volume in the ith pixel. It may therefore be sur-
prising at first glance to find out that the average mean transit time, Tm ,
cannot be calculated, similar to equation (1a) and (1b), as:
1 N
Tm = ∑T
N i=1 m,i
(2)
where Tm,i is the mean transit time through the ith pixel. Consider a ROI
consisting of two pixels. In one pixel blood flow is zero, but blood
volume can be either zero or non-zero; by definition the mean transit
time in this pixel is zero. In the second pixel, blood flow, blood
volume, and mean transit time are all finite. In this case, the mean tran-
sit time of the ROI should be equal to the finite mean transit time of the
2nd pixel, Tm,2; however, according to equation (2), it is equal to one-
half of Tm,2 instead, which is incorrect. The proper definition of the
average mean transit time of a ROI should be:
N N
Fi
Tm = ∑ T where FT = ∑ Fi (3)
i=1 FT m,i i=1
or, the average mean transit time of a ROI is the flow weighted
average of mean transit time of each pixel within the ROI. Since
1 N
1 N Vb,T Vb
Tm =
FT
∑F ⋅T
i=1
i m,i
=
FT
∑V
i=1
b,i
=
FT
=
F
(5)
Equation (4) states that the central volume principle applies to average
values of blood flow, blood volume, and mean transit time from a ROI
in the same way as it applies to values from single pixels, provided that
the average mean transit time of a ROI is defined as in equation (3).
IMAGE PROCESSING
a b
c d
Image processing 67
a b
c d
Figure 4.5 (a) Contrast-enhanced CT and perfusion images (b) without seg-
mentation, (c) with elimination of non-tumor tissues, and (d) with elimination
of non-tumor tissues and vascular structures. Non-tumor tissues were removed
by excluding pixels with attenuation values before administration of contrast
medium of less than 0 HU and more than 100 HU. Vascular structures were
eliminated by removing pixels with peak enhancement more than 10% of the
peak enhancement within the aorta
perfusion imaging of the brain, Kudo et al.2 eliminated pixels with per-
fusion values above 8 ml min−1100 g−1, whilst Sase et al.3 defined crite-
ria based on ratios of peak enhancement values within pixels to the
peak enhancement within the superior sagittal sinus. Similar processes
can be applied to non-cerebral tumors located close to large or
medium-sized blood vessels (Figure 4.5d). Elimination of vascular
pixels in this way improves the accuracy of perfusion measurements.2
Image processing 69
a b
720 720
0 0
c d
90 5
0 0
Figure 4.6 (a) Perfusion-weighted image of the same patient as in Figure 4.1
with five regions of interest (ROIs) drawn roughly to outline the whole brain (1),
the intercerebral fissure (2), the primary tumor (3), and two metastases (4 and 5);
(b) thresholding to highlight gray matter () and white matter () in normal
tissue and enhancing () and non-enhancing regions () in tumor; (c) the
pixel masks of gray and white matter as well as enhancing and non-enhancing
tumor regions applied to the blood flow map to determine the average blood
flow in gray matter or enhancing tumor regions () and white matter or non-
enhancing tumor regions (); (d) the pixel masks of gray and white matter
as well as enhancing and non-enhancing tumor regions applied to the
blood volume map to determine the average blood volume in gray matter or
enhancing tumor regions () and white matter or non-enhancing tumor
regions ()
Table 4.3 Average blood flow in the different tissue types identified by automatic
thresholding in the regions of interest (ROIs) shown in Figure 4.6a
CONCLUSION
Table 4.4 Average blood volume in the different tissue types identified by automatic
thresholding in the ROIs shown in Figure 4.6a
Image processing 71
Table 4.5 Average capillary permeability surface area product (PS) in the different tissue
types identified by automatic thresholding in the ROIs shown in Figure 4.6a
PS ml/min/100 g
Tissue type Average SD Pixel count
Table 4.6 Comparison of average mean transit time (equation (2)) and flow weighted
average mean transit time (equation (5)) in the different tissue types identified by
automatic thresholding in the ROIs shown in Figure 4.6a
REFERENCES
5
Angiogenesis, tumor perfusion,
and cancer management
William W Li, Aliya Jiwani, and Michelle Hutnik
INTRODUCTION
The incipient stage of all solid tumors is growth restriction due to their
intrinsic lack of a supportive blood supply.1,3,4 At the 60–80-cell stage,
such tumors often grow marginally towards nearby blood vessels
through a process termed vessel cooption.5 Some malignant cells may
infiltrate these local blood vessels to form a ‘mosaic’ vessel comprising
normal vascular cells interspersed with a few tumor cells.6 Vascular
cooption serves the tumor periphery, so tumor expansion leads to
increasing central hypoxia, limiting tumor size to approximately
2–5 millimeters in diameter, until neovascularization takes place. This
initial ‘prevascular’ phase of limited tumor growth may persist for years
without clinical signs or detection.7,8
The rapid phase of tumor growth occurs following the initiation of
tumor angiogenesis, a phenomenon known as the ‘angiogenic switch’,
in which a subset of cancer cells begins secreting angiogenic stimula-
tory factors in excess of inhibitors, thus activating local endothelial cells
to proliferate.9,10 Both human and experimental cancers produce
9781842143094-Ch05 8/8/07 11:56 AM Page 75
a b
in the center of the tumor.29 In general, blood flow in tumor vessels can
be variable and sluggish.29 It can temporarily stop or even reverse direc-
tion in sections of the tumor.27 Consequently, the tumor vasculature is
inefficient in the delivery of nutrients and oxygen to hypoxic regions,
leading to hypoxia-induced generation of angiogenic factors (i.e. VEGF)
and further rounds of tumor angiogenesis.27
IMAGING ANGIOGENESIS
The small size (< 100 µm) of tumor microvessels precludes direct
visualization with conventional angiography.26 Current clinical imaging
modalities provide much higher-resolution images than microscopic
methods originally used to study tumor vasculature.27 The combination
of high-resolution imaging with the ability to monitor dynamic (func-
tional) processes has produced major advances in observing and
measuring tumor vasculature. The optimization, validation, and stan-
dardization of these angiogenesis imaging procedures are the next steps
required to accurately compare and consistently reproduce data.46
Coordinated efforts between radiologists, nuclear medicine specialists,
surgeons, medical oncologists, radiation oncologists, and pathologists
will be paramount to the understanding, acceptance, and application of
new imaging techniques.47
Imaging modalities that have been used and are in development to
study vascular features in tumor include high-resolution angiography,
functional computer tomography (CT) imaging, dynamic contrast-
enhanced magnetic resonance imaging (DCE-MRI), high-resolution
magnetic resonance angiography (MRA), positron emission tomography
(PET), Doppler ultrasound (US), and contrast-enhanced US.48,49
Functional CT
CT imaging can be performed with contrast agents to define the
intravascular compartment, including blood flow, blood volume, tissue
perfusion, capillary permeability, and leakage.48,50 Functional CT tech-
niques can delineate increases in tissue perfusion that may reflect
malignancy, even when there is no gross anatomical abnormality
present.50–52
DCE-MRI
MRI can define both blood volume and blood vessel permeability by
using dynamic enhancement of blood pool contrast agents.53,54 The
use of gadolinium can assist in distinguishing between normal (physi-
ological permeability) and malignant (hyperpermeable) tissues,
although the rapid exchange of water between plasma and cell
membranes can lead to an overestimation of the volume of the tissue
compartment by contrast enhancement. Contrast material uptake also
correlates with microvessel density in experimental tumors.
Administration of an anti-VEGF monoclonal antibody to experimental
breast cancers in mice causes decreased vascular permeability that is
detectable with MR imaging.27,55
9781842143094-Ch05 8/8/07 11:56 AM Page 80
MRA
Whereas DCE-MRI focuses on the tumor microvasculature, high-resolution
MRA can be used to view the macroscopic tumor vascular network,
specifically the morphology of larger tumor vessels in the brain.56–59
Initial studies of abnormalities of the larger vessel shape, or tortuosity,
have found a correlation between tumor vessel tortuosity and whether
the tumor is benign or malignant.58
PET
PET imaging is used to evaluate tumor metabolism, as well as blood
flow and volume.60,61 A number of radiotracers, such as water labelled
with oxygen-15, carbon monoxide with carbon-11, and fluorodeoxyglu-
cose with fluorine-18 (18FDG), are available to characterize neoplastic
tissue. By interfering with tumor blood flow, antiangiogenic agents
decrease tumor metabolism. Studies using radiolabeled fluoromisonida-
zole together with PET have been carried out to quantify hypoxia in the
rat glioma and provide functional information about the results of
antiangiogenic therapy.62 A number of novel tracers targeting vascular
integrins have shown experimental utility in characterizing tumor
angiogenesis using PET imaging.63–65
Ultrasound
Ultrasound (US) imaging can identify vascular features in tumors at dif-
ferent levels of resolution (40–200-µm diameter vessels), depending on
the technique employed.66 Contrast agents have dramatically improved
the spatial resolution of US.48 The microbubbles used as contrast agents
for US are generally intravascular and not interstitial, and therefore an
index of blood flow, blood volume, or vascularity within malignant
tissue can be generated. Color flow Doppler US, with or without con-
trast agents, has been used to characterize tumor xenografts in mice
and solid tumors in patients.67–69
CONCLUSIONS
ACKNOWLEDGMENTS
REFERENCES
6
Tumors of the brain,
head, and neck
INTRODUCTION
MENINGIOMA
a b 500 ml/min/100 ml
0 ml/min/ml
c d 50 ml/min/100 ml
20 ml/100 ml
0 ml/100 ml 0 ml/min/100 ml
Table 6.1 Computed tomography (CT) perfusion values for normal cerebral tissues and
meningioma as reported by Cheong et al. using compartmental analysis17
GLIOMA
a b
High grade gliomas are usually treated with adjuvant radiation therapy
or chemotherapy after resection, whereas this is not the case for low
grade gliomas.6 Conventional contrast-enhanced CT can provide only a
limited amount of information concerning the grade of glioma. Thus,
the current standard for tumor grading is histopathological assessment,
but this is complicated by the heterogeneous nature of gliomas, and his-
tological samples obtained at biopsy may be subject to sampling error.8
Hence, there is a need for a more accurate and less invasive method of
tumor grading.
In a study of 22 patients, Ding et al. assessed the ability of CT perfusion
to grade cerebral glioma.21 Both CBV and PS were strongly correlated
with tumor grade, and the differences in CBV and PS were statistically
significant between low- and high-grade glioma.21 In addition, receiver
operating characteristic curves revealed better diagnostic performance for
PS in comparison to CBV for determining glioma grade.21 These findings
are analogous to those reported previously using MR.1,5,6,8
CEREBRAL LYMPHOMA
on conventional CT can also differ from other tumors, with most cases
showing homogeneous contrast enhancement. Warnke et al. have used
CT perfusion to study seven patients with cerebral lymphoma.24 As in
the case of other tumors, vascular permeability was markedly increased
compared to normal cerebral tissue (2.95 ± 1.06 ml/min/100 g), and
higher than in other cerebral tumors assessed using the same technique.
However, the relative blood volume of tumor was comparable to that of
normal brain (2.7 ± 2.4 ml/100 g), and tumor perfusion values, measured
in this study by xenon CT, were also close to those for normal white
matter (43.2 ± 10.5 ml/min/100 g).
CT CBV CBF
Cerebral tumors typically exhibit CBF and CBV values that are similar to
or higher than those obtained from normal brain tissue. In contrast, non-
hemorrhagic stroke is readily identified on CT perfusion as an area of
reduced CBF.29 CBV may be normal in the presence of reversible ischemia,
reflecting preservation of vascular autoregulation, whereas irreversible infarc-
tion is characterized by matched reductions in CBF and CBV.3 However,
images of vascular permeability are less likely to differentiate cerebral tumor
and infarction, as values are increased above normal in both pathologies.
Radiation planning
The management of malignant cerebral neoplasms requires aggressive
treatment modalities including surgery and radiotherapy, but the prognosis
of these tumors still remains poor. In order to maximize the radiation
dose to the tumor and minimize the damage to normal surrounding
tissue, an accurate and reliable identification of viable tissue margins
needs to be obtained.31 Anatomic images may not sufficiently identify
eloquent cortex, and may misjudge the distance from potential treatment
margins. Incorporation of physiological imaging into the planning
process may improve tumor targeting by more accurate target definition
and reduction of the total target volume. Furthermore, there is interest in
9781842143094-Ch06 8/8/07 4:24 PM Page 98
a b
THERAPEUTIC MONITORING
CONCLUSION
INTRODUCTION
The aim of a study reported by Hermans et al.47 was to test the hypoth-
esis that the outcome of patients is affected by the tumor perfusion rate.
Multivariate analysis confirmed the value of CT-determined tumor per-
fusion as an independent predictor of local control in head and neck
cancer, treated by definitive radiotherapy, with or without adjuvant
chemotherapy. Patients with a low perfusion value showed a statistically
significantly higher local failure rate than those with a high perfusion
value (Figure 6.6). Presumably, this is linked to more extensive and/or a
higher degree of hypoxia in low-perfused tumors, as suggested by others.54
The perfusion rate was also found to be independent of tumor
volume. CT-determined primary tumor volume has been shown to be
an important predictor of local control for several head and neck cancer
sites, including glottic,55 supraglottic,56 hypopharyngeal,57 and nasopha-
ryngeal cancer.58,59 As perfusion rate is independent of tumor volume,
it can be considered an additional parameter helping to predict the out-
come of the patient after treatment.
The perfusion rate was found to be independent of the T-classification.47
Local control was significantly different according to the median perfu-
sion rate in the subgroups T3 and T4 (Figures 6.7 and 6.8). Particularly
in the T4-group, the difference in local outcome between those patients
with high and low perfusion values is very pronounced. This opens
perspectives to use CT perfusion as a tool to select patients suffering
advanced head and neck cancer who may benefit from concomitant
treatment during irradiation.
The practical advantages of this method are the low extra burden to
patients, the ease of implementing it on existing CT machines, and
9781842143094-Ch06 8/8/07 4:24 PM Page 103
1
< 83.5 ml/min/100 g
0.9
> 83.5 ml/min/100 g
0.8
0.7
Local control
0.6
0.5
0.4
0.3
0.2
0
0 6 12 18 24 30 36 42
Time (months)
Figure 6.6 Local control over time in 105 patients suffering head and neck
cancer. The patients are stratified into two groups according to the median
primary tumor perfusion value, as determined by dynamic CT using the
gradient method. (Reproduced with permission from reference 47)
1
< 83.5 ml/min/100 g
0.9
> 83.5 ml/min/100 g
0.8
0.7
Local control
0.6
0.5
0.4
0.3
0.2
0
0 6 12 18 24
Time (months)
Figure 6.7 Local control versus perfusion rate (patients classified as T3 (n = 39))
plotted over time after start of radiotherapy. (Reproduced with permission from
reference 47)
9781842143094-Ch06 8/8/07 4:24 PM Page 104
1
< 83.5 ml/min/100 g
0.9
> 83.5 ml/min/100 g
0.8
0.7
Local control
0.6
0.5
0.4
0.3
0.2
0
0 6 12 18 24
Time (months)
Figure 6.8 Local control versus perfusion rate (patients classified as T4 (n = 34))
plotted over time after start of radiotherapy. (Reproduced with permission from
reference 47)
Figure 6.9 Time frame of a dynamic series obtained during the intravenous
injection of a contrast agent bolus, using a four-row multidectector CT machine.
A pixel-by-pixel analysis of the dynamic image series was performed. The perfusion
data are color-coded and mapped on the anatomical images. The perfusion rate
shows a heterogeneous distribution throughout the tumor (calculated using
Body CT Perfusion; Siemens Medical Solutions, Erlangen, Germany)
CONCLUSION
burden for the patient. This parameter has significant predictive value
for local outcome after irradiation with curative intent, and therefore
may be considered a useful tool to select patients for concomitant treat-
ment in advanced head and neck cancer.
REFERENCES
1. Roberts HC, Roberts PL, Brasch RC, Dillon WP. Quantitative measurements of
microvascular permeability in human brain tumors achieved using dynamic
contrast-enhanced MR imaging: correlation with histologic grade. AJNR Am J
Neuroradiol 2000; 21: 891–9.
2. Leggett DAC, Miles KA, Kelley BB. Blood-brain barrier and blood volume
imaging of cerebral glioma using functional CT: a pictorial review. Eur J Radiol
1999; 30: 185–90.
3. Miles KA. Brain perfusion: computed tomography applications. Neuroradiology
2004; 46: S194–S200.
4. Miles KA, Eastwood JD, Konig M, eds. Multidetector Computed Tomography
in Cerebrovascular Disease: CT Perfusion Imaging. Abingdon, UK: informa
Healthcare, 2007.
5. Jackson A, Kassner A, Annesley-Williams D et al. Abnormalities in the recircula-
tion phase of contrast agent bolus passage in cerebral gliomas: comparison with
relative blood volume and tumor grade. AJNR Am J Neuroradiol 2002; 23: 7–14.
6. Law M, Yang S, Babb JS et al. Comparison of cerebral blood volume and
vascular permeability from dynamic susceptibility contrast-enhanced perfusion
MR imaging with glioma grade. AJNR Am J Neuroradiol 2004; 25: 746–55.
7. Provenzale JM, Mukundan S., Dewhirst M. The role of blood-brain barrier
permeability in brain tumor imaging and therapeutics. AJR Am J Roentgenol
2005; 185: 763–7.
8. Aronen HJ, Gazit IE, Louis DN et al. Cerebral blood volume maps of gliomas:
comparison with tumor grade and histologic findings. Radiology 1994; 191: 41–51.
9. Miles, KA, Charnsangavej C, Lee FT et al. Application of CT in the investiga-
tion of angiogenesis in oncology. Acad Radiol 2000; 7: 840–50.
10. Miles KA, Wintermark M. Comparison with other techniques. In: Miles KA,
Eastwood JD, Konig M, eds. Multidetector Computed Tomography in
Cerebrovascular Disease: CT Perfusion Imaging. Abingdon, UK: Informa
Healthcare, 2007; 159–70.
11. Rachlin JR. Etiology and biology of meningiomas. In: Al-Mefty O, ed.
Meningiomas. New York: Raven Press, 1991; 27–35.
12. Russel DS, Rubenstein LJ. Pathology of Tumors of the Central Nervous System,
4th edn. London: Edward Arnold, 1995: 66–91.
13. Buetow MP, Buetow PC, Smirniotopoulos JG. Typical, atypical and misleading
features in meningioma. Radiographics 1991; 11: 1087–106.
14. Hakyemez B, Yildirim N, Gokalp G, Erdogan C, Parlak M. The contribution of
diffusion-weighted MR imaging to distinguish typical from atypical meningiomas.
Neuroradiology 2006; 48: 513–20.
15. Jinkins JR, Nuri Sener R. The characteristics of cerebral meningiomas and
surrounding tissues on dynamic CT. Neuroradiology 1991; 33: 499–506.
9781842143094-Ch06 8/8/07 4:24 PM Page 107
33. Jain R, Elika S, Searpace L, Schultz L, Mikkelsen T, Pakel S. Perfusion CT: Initial
experience in differentiating recurrent brain tumors from radiation effects/
radiation necrosis. European Radiology 2007; 17: 291.
34. Sugahara T, Korogi Y, Tomiguchi S et al. Posttherapeutic intraaxial brain
tumor: the value of perfusion-sensitive contrast-enhanced MR imaging for
differentiating tumor recurrence from non-neoplastic contrast-enhancing
tissue. Am J Neuroradiol 2000; 21: 901–9.
35. Gray LH, Conger AD, Ebert M et al. The concentration of oxygen dissolved in
tissues at the time of irradiation as a factor in radiotherapy. Br J Radiol 1953;
26: 638–48.
36. Vaupel P, Kelleher DK, Hockel M. Oxygen status of malignant tumors: patho-
genesis of hypoxia and significance for tumor therapy. Semin Oncol 2001; 28
(Suppl 8): S29–35.
37. Dunst J, Stadler P, Becker A et al. Tumor volume and tumor hypoxia in head
and neck cancers. The amount of the hypoxic volume is important. Strahlenther
Onkol 2003; 179: 521–6.
38. Brizel DM, Dodge RK, Clough RW et al. Oxygenation of head and neck cancer:
changes during radiotherapy and impact on treatment outcome. Radiother
Oncol 1999; 53: 113–17.
39. Nordsmark M, Overgaard J. Tumor hypoxia is independent of hemoglobin and
prognostic for loco-regional tumor control after primary radiotherapy in
advanced head and neck cancer. Acta Oncol 2004; 43: 396–406.
40. Wouters BG, Weppler SA, Koritzinsky M et al. Hypoxia as a target for com-
bined modality treatments. Eur J Cancer 2002; 38: 240–57.
41. Martin L, Lartigau E, Weeger P et al. Changes in the oxygenation of head and
neck tumors during carbogen breathing. Radiother Oncol 1993; 27: 123–30.
42. Becker A, Kuhnt T, Liedtke H et al. Oxygenation measurements in head and
neck cancers during hyperbaric oxygenation. Strahlenther Onkol 2002; 178:
105–8.
43. Kaanders JH, Pop LA, Marres HA et al. Accelerated radiotherapy with
carbogen and nicotinamide (ARCON) for laryngeal cancer. Radiother Oncol
1998; 48: 115–22.
44. Kaanders JH, Pop LA, Marres HA et al. ARCON: experience in 215 patients
with advanced head-and-neck cancer. Int J Radiat Oncol Biol Phys 2002; 52:
769–78.
45. Hermans R, Lambin P, Van den Bogaert et al. Non-invasive tumor perfusion
measurement by dynamic CT: preliminary results. Radiother Oncol 1997; 44:
159–62.
46. Hermans R., Lambin P, Van der Goten A et al. Tumoral perfusion as measured
by dynamic computed tomography in head and neck carcinoma. Radiother
Oncol 1999; 53: 105–11.
47. Hermans R, Meijerink M, Van den Bogaert W et al. Tumor perfusion rate deter-
mined non-invasively by dynamic computed tomography predicts outcome in
head and neck cancer after radiotherapy. Int J Radiat Oncol Biol Phys 2003;
57: 1351–6.
48. Mullani NA, Gould KL. First-pass measurements of regional blood flow with
external detectors. J Nucl Med 1983; 24: 577–81.
49. Blomley MJK, Coulden R, Bufkin C et al. Contrast bolus dynamic computed
tomography for the measurement of solid organ perfusion. Invest Radiol 1993;
28: S72–7.
9781842143094-Ch06 8/8/07 4:24 PM Page 109
50. Blomley MJK, Coulden R, Dawson P et al. Liver perfusion studied with ultra-
fast CT. J Comput Assist Tomogr 1995; 19: 424–33.
51. Brix G, Bahner ML, Hoffmann U et al. Regional blood flow, capillary perme-
ability, and compartmental volumes: measurement with dynamic CT – initial
experience. Radiology 1999; 210: 269–76.
52. Jaschke W, Gould RG, Assimakopoulos PA et al. Flow measurements with a
high-speed computed tomography scanner. Med Phys 1987; 14: 238–43.
53. Gandhi D, Hoeffner EG, Carlos RC et al. Computed tomography perfusion of
squamous cell carcinoma of the upper aerodigestive tract. Initial results.
J Comput Assist Tomogr 2003; 27:687–93.
54. Lartigau E, Le Ridant AM, Lambin P et al. Oxygenation of head and neck
tumors. Cancer 1993; 71:2319–25.
55. Hermans R, Van den Bogaert W, Rijnders A et al. Predicting the local outcome
of glottic cancer treated by definitive radiation therapy: value of computed
tomography determined tumor parameters. Radiother Oncol 1999; 50: 39–46.
56. Mancuso AA, Mukherji SK, Schmalfuss I et al. Preradiotherapy computed
tomography as a predictor of local control in supraglottic carcinoma. J Clin
Oncol 1999; 17: 631–7.
57. Pameijer FA, Mancuso AA, Mendenhall WM et al. Evaluation of pretreatment
computed tomography as a predictor of local control in T1/T2 pyriform sinus
carcinoma treated with definitive radiotherapy. Head Neck 1998; 20: 159–68.
58. Chua DT, Sham JS, Kwong DL et al. Volumetric analysis of tumor extent in
nasopharyngeal carcinoma and correlation with treatment outcome. Int J
Radiat Oncology Biol Phys 1997; 39: 711–19.
59. Kim JH, Lee JK. Prognostic value of tumor volume in nasopharyngeal carcinoma.
Yonsei Med J 2005; 30: 221–7.
60. Chao KSC, Bosch WR, Mutic S et al. A novel approach to overcome hypoxic
tumor resistance: Cu-ATSM-guided intensity-modulated radiation therapy. Int J
Radiat Oncol Biol Phys 2001; 49: 1171–82.
61. Römer W, Muresan L, Adamietz B et al. Biological characterization and response
monitoring in head and neck cancer therapy using multislice perfusion CT.
Eur Radiol 2002; 12 (Suppl 1): S193 (abstr).
62. Million RR, Cassisi NJ, Mancuso AA et al. Management of the neck for squa-
mous cell carcinoma. In: Million RR, Cassisi NJ, eds. Management of Head
and Neck Cancer: a Multidisciplinary Approach. Philadelphia, PA: JB
Lippincott Company, 1994: 125.
9781842143094-Ch06 8/8/07 4:24 PM Page 110
9781842143094-Ch07 8/8/07 11:57 AM Page 111
7
CT perfusion applications
in lung cancer
Kwun M Fong, Rayleen V Bowman, Ian A Yang, and Kenneth A Miles
INTRODUCTION
There are a large number of malignant tumors that affect the lungs,
ranging from metastatic malignancies to the most common type, bron-
chogenic carcinoma. Bronchogenic cancers can also metastasize to the
lungs, as well as other sites; but the lung is also a common destination
for the metastatic spread of a wide variety of other human malignancies.
Bronchogenic carcinoma (lung cancer) is the highest ranking cause
of cancer-associated mortality in the world. Historically more prevalent
in men, bronchogenic carcinoma is currently causing higher mortality
among women living in several Western countries than is breast or any
other cancer. Non-small-cell lung cancer accounts for approximately
80–85% of primary lung cancer, and small-cell cancer accounts for most
of the remainder. The two types differ in both biological characteristics
and clinical behavior.
Presentation
Lung cancers can present in a large number of ways. A frequently
encountered clinical situation is the presentation of a patient with or
without symptoms who is found to have a pulmonary nodule on chest
radiography or on thoracic CT undertaken for another reason, such as
diagnosis of suspected pulmonary embolism. For this common clinical
scenario where lung cancer is a diagnostic possibility, the challenge is
first to ascertain the presence or absence of malignancy. If malignancy
is confirmed, the clinician has to determine the origin of the cancer, and
finally the extent or stage of disease so that therapy may be appropri-
ately individualized. Therefore, tools to help decide the likelihood of
malignancy and the subsequent need for invasive investigations for
pulmonary lesions are essential for the clinician who manages patients
with suspected lung cancer.
by large screening studies in some parts of the world, which show non-
calcified nodules in the majority of screenees.1 Therefore, if low-dose
CT comes into routine use, there will be an exponential increase in the
number of nodules that require clinical evaluation.
As most identified pulmonary nodules (defined as an intraparenchy-
mal lung lesion less than 3 cm in diameter and not associated with atelec-
tasis or lymphadenopthy2 turn out not to be malignant, there is a pressing
need for strategies to efficiently triage pulmonary nodules so that those
requiring treatment can be identified, while avoiding the expense, risk,
and inconvenience of unnecessary invasive investigations.
CONTRAST-ENHANCED CT
With the advent of spiral CT, there has been a rapid profusion of
research protocols aimed at distinguishing differences in vascularity
between malignant and benign tumors. In addition, there have also
been a variety of measurements utilized for quantifying the results from
contrast-enhanced nodule scanning.
Simple measurements of tissue attenuation9 or enhancement10 can be
obtained, as can absolute measures of tissue perfusion.11 Peak enhance-
ment is a measure of the maximum increase in density after contrast
administration, and is determined by not only tumor perfusion but also
blood volume and capillary permeability. The degree to which peak
enhancement approximates perfusion is dependent upon the transit time
of contrast material through the tumor circulation: the longer is the tran-
sit time, the closer is the dependence of peak enhancement on perfusion.
9781842143094-Ch07 8/8/07 11:57 AM Page 114
Box 7.1 Technique for quantifying contrast enhancement within lung nodules: repeated
spiral (helical) acquisitions (based on reference 13)
a b c
Enhancement = 30 HU
a
T
1 5
2 4
Figure 7.2 Phantom for calibrating computed tomography (CT) systems for
sensitivity to iodine. (a) End plan-view and (b) overall view. Contrast material
at various dilutions is introduced into cylinders of diameter 25 mm and length
50 mm located at positions 1–5 as follows: (1) 1:50, (2) 1:35, (3) 1:20,
(4) 1:200, (5) 1:100. Three small venting plugs (V) allow release of air when filling
the phantom with water. (Reproduced with permission from reference 14)
Box 7.2 Protocol for generating consistent CT images of peak enhancement within
lung masses
Aim: To generate images of peak enhancement within lung nodules whilst minimizing
inconsistencies due to motion and/or segmentation of the tumor. The resulting images
will allow assessment of regional variations in tumor peak enhancement.
1. Create a region of interest (ROI) encompassing the pulmonary nodule displayed with
mediastinal windows.
2. Use image thresholds of −50 HU to 100 HU to eliminate from analysis any pixels
containing air or calcium.
3. Exclude large blood vessels from the ROI by eliminating pixels that are enhanced by
more than 100 HU following contrast material.
4. Display the changes in X-ray attenuation within the ROI following administration of
contrast material as a time-attenuation curve and identify the time of maximal tumor
enhancement. Only the image at this time point and the baseline image are retained
for further analysis.
5. If several images have been acquired at each time, compensate for respiratory motion
by selecting from each data set the images that correspond most closely in anatomical
position.
6. Generate parametric image displaying peak tumor enhancement (and/or standardized
perfusion value) by subtracting the baseline image from the image in which the
attenuation within tumor ROI was maximal.
Indeterminate
Nodule 2 year CT
Low
enhanced follow-up for
enhancement
CT stability
High enhancement
FDG Negative
PET
Positive
Biopsy/
resection
POTENTIAL LIMITATIONS
Variations in technique
To date, there does not appear to be any consensus as to a standard-
ized protocol, including the amount of contrast material and flow rate
to be used, the frequency of data recording, the scanning time, and
the scan interval. Each of these factors has the potential to affect the
pattern and quantity of measurable enhancement.28
Small nodules
Most if not all studies report contrast enhancement in pulmonary nodules
of at least 5 or 6 mm in diameter.3,16,23 There is limited knowledge of the
value of this technique in very small lesions.
Tumoral heterogeneity
Yamashita et al. noted that some lung cancers demonstrate inhomo-
geneous patterns of contrast enhancement at CT, some correlating
to necrosis, fibrosis, or perhaps differentiation.21 This implies that
necrotic cancers may not demonstrate the high levels of enhancement
seen with non-necrotic tumors (Figure 7.4).24 Indeed some experts
recommend that the use of contrast-enhanced CT be restricted to
pulmonary nodules with diameters 2–2.5 cm or less, as they are less
likely to have substantial necrosis, and thus be falsely negative.3,26
Furthermore, the clinical impact of smaller lesions is different, as they
are more likely to be benign, and are generally more difficult to
biopsy successfully either bronchoscopically or via a transthoracic
percutaneous approach.
9781842143094-Ch07 8/8/07 11:57 AM Page 123
a b 50HU
0HU
Figure 7.4 Large non-small-cell lung cancer with low-level and heterogeneous
enhancement implying necrosis. (a) Conventional contrast-enhanced CT;
(b) maximum-enhancement image.
Radiation dose
Many contrast CT scans are utilizing tube voltages of 80 kV in an attempt
to reduce radiation exposure. This voltage also has the advantage of
increasing the sensitivity to iodine-based contrast agents. However, for
measures of nodule enhancement, new diagnostic thresholds will need
to be established for reduced tube voltages. It has been estimated using
International Commission on Radiation Protection (ICRP) recommenda-
tions and the CT-Expo program for CT dose evaluation that there is an
effective dose of about 1.3 mSv.22
FUTURE DEVELOPMENTS
Refinements
Heterogeneous enhancement is more commonly a feature of malignant
nodules.18 Although heterogeneity can be assessed visually, this image
9781842143094-Ch07 8/8/07 11:57 AM Page 124
CONCLUSIONS
REFERENCES
1. Swensen SJ, Jett JR, Sloan JA et al. Screening for lung cancer with low-dose
spiral computed tomography. Am J Respir Crit Care Med 2002; 165: 508–13.
2. Tuddenham WJ. Glossary of terms for thoracic radiology: recommendations of
the Nomenclature Committee of the Fleischner Society. AJR Am J Roentgenol
1984; 143: 509–17.
3. Swensen SJ. Functional CT: lung nodule evaluation. Radiographics 2000; 20:
1178–81.
4. Viamonte M Jr. Angiographic evaluation of lung neoplasms. Radiol Clin North
Am 1965; 3: 529–42.
5. Littleton JT, Durizch ML, Moeller G, Herbert DE. Pulmonary masses: contrast
enhancement. Radiology 1990; 177: 861–71.
6. Patz EF Jr, Lowe VJ, Hoffman JM et al. Focal pulmonary abnormalities:
evaluation with F-18 fluorodeoxyglucose PET scanning. Radiology 1993; 188:
487–90.
7. Yuan A, Chang DB, Yu CJ et al. Color Doppler sonography of benign and
malignant pulmonary masses. AJR Am J Roentgenol 1994; 163: 545–9.
8. Guckel C, Schnabel K, Deimling M, Steinbrich W. Solitary pulmonary nodules:
MR evaluation of enhancement patterns with contrast-enhanced dynamic
snapshot gradient-echo imaging. Radiology 1996; 200: 681–6.
9. Yamashita K, Matsunobe S, Tsuda T et al. Solitary pulmonary nodule: prelimi-
nary study of evaluation with incremental dynamic CT. Radiology 1995; 194:
399–405.
9781842143094-Ch07 8/8/07 11:57 AM Page 127
10. Swensen SJ, Brown LR, Colby TV, Weaver AL, Midthun DE. Lung nodule
enhancement at CT: prospective findings. Radiology 1996; 201: 447–55.
11. Miles KA, Griffiths MR, Fuentes MA. Standardized perfusion value: universal
CT contrast enhancement scale that correlates with FDG PET in lung nodules.
Radiology 2001; 220: 548–53.
12. Jeong YJ, Lee KS, Jeong SY et al. Solitary pulmonary nodule: characterization
with combined wash-in and washout features at dynamic multi-detector row
CT. Radiology 2005; 237: 675–83.
13. Swensen SJ, Viggiano RW, Midthun DE et al. Lung nodule enhancement at CT:
multicenter study. Radiology 2000; 214: 73–80.
14. Miles KA, Young H, Chica SL, Esser PD. Quantitative contrast-enhanced com-
puted tomography: is there a need for system calibration? Eur Radiol 2007; 17:
919–26.
15. Erasmus JJ, Connolly JE, McAdams HP, Roggli VL. Solitary pulmonary nodules:
Part I. Morphologic evaluation for differentiation of benign and malignant
lesions. Radiographics 2000; 20: 43–58.
16. Swensen SJ, Morin RL, Schueler BA et al. Solitary pulmonary nodule: CT eval-
uation of enhancement with iodinated contrast material—a preliminary report.
Radiology 1992; 182: 343–7.
17. Zhang M, Kono M. Solitary pulmonary nodules: evaluation of blood flow
patterns with dynamic CT. Radiology 1997; 205: 471–8.
18. Petkovska I, Shah SK, McNitt-Gray MF et al. Pulmonary nodule characterization:
a comparison of conventional with quantitative and visual semi-quantitative
analyses using contrast enhancement maps. Eur J Radiol 2006; 59: 244–52.
19. Yi CA, Lee KS, Kim BT et al. Tissue characterization of solitary pulmonary
nodule: comparative study between helical dynamic CT and integrated
PET/CT. J Nucl Med 2006; 47: 443–50.
20. Christensen JA, Nathan MA, Mullan BP et al. Characterization of the solitary
pulmonary nodule: 18F-FDG PET versus nodule-enhancement CT. AJR Am J
Roentgenol 2006; 187: 1361–7.
21. Yamashita K, Matsunobe S, Takahashi R et al. Small peripheral lung carcinoma
evaluated with incremental dynamic CT: radiologic-pathologic correlation.
Radiology 1995; 196: 401–8.
22. Kiessling F, Boese J, Corvinus C et al. Perfusion CT in patients with advanced
bronchial carcinomas: a novel chance for characterization and treatment
monitoring? Eur Radiol 2004; 14: 1226–33.
23. Yi CA, Lee KS, Kim EA et al. Solitary pulmonary nodules: dynamic enhanced
multi-detector row CT study and comparison with vascular endothelial growth
factor and microvessel density. Radiology 2004; 233: 191–9.
24. Yamashita K, Matsunobe S, Tsuda T et al. Intratumoral necrosis of lung
carcinoma: a potential diagnostic pitfall in incremental dynamic computed
tomography analysis of solitary pulmonary nodules? J Thorac Imaging 1997;
12: 181–7.
25. Tateishi U, Nishihara H, Tsukamoto E et al. Lung tumors evaluated with FDG-
PET and dynamic CT: the relationship between vascular density and glucose
metabolism. J Comput Assist Tomogr 2002; 26: 185–90.
26. Miles KA, Griffiths MR, Keith CJ. Blood flow-metabolic relationships are depend-
ent on tumour size in non-small cell lung cancer: a study using quantitative
contrast-enhanced computer tomography and positron emission tomography.
Eur J Nucl Med Mol Imaging 2006; 33: 22–8.
9781842143094-Ch07 8/8/07 11:57 AM Page 128
27. Comber LA, Keith CJ, Griffiths M, Miles KA. Solitary pulmonary nodules:
impact of quantitative contrast-enhanced CT on the cost-effectiveness of
FDG-PET. Clin Radiol 2003; 58: 706–11.
28. Marten K, Grabbe E. The challenge of the solitary pulmonary nodule: diagnostic
assessment with multislice spiral CT. Clin Imaging 2003; 27: 156–61.
29. Diederich S, Theegarten D, Stamatis G, Luthen R. Solitary pulmonary nodule
with growth and contrast enhancement at CT: inflammatory pseudotumour as
an unusual benign cause. Br J Radiol 2006; 79: 76–8.
30. Matsuki M, Noma S, Kuroda Y et al. Thin-section CT features of intrapul-
monary lymph nodes. J Comput Assist Tomogr 2001; 25: 753–6.
31. Shah SK, McNitt-Gray MF, Rogers SR et al. Computer aided characterization of
the solitary pulmonary nodule using volumetric and contrast enhancement
features. Acad Radiol 2005; 12: 1310–19.
32. Meert AP, Paesmans M, Martin B et al. The role of microvessel density on the
survival of patients with lung cancer: a systematic review of the literature with
meta-analysis. Br J Cancer 2002; 87: 694–701.
33. Shim SS, Lee KS, Chung MJ et al. Do hemodynamic studies of stage T1 lung
cancer enable the prediction of hilar or mediastinal nodal metastasis? AJR Am
J Roentgenol 2006; 186: 981–8.
34. Choi J-B, Park C-K, Park DW et al. Does contrast enhancement on CT suggest
tumor response for chemotherapy in small cell carcinoma of the lung?
J Comput Assist Tomogr 2002; 26: 797–800.
35. Ng QS, Goh V, Fichte H et al. Lung cancer perfusion at multi-detector row CT:
reproducibility of whole tumor quantitative measurements. Radiology 2006;
239: 547–53.
36. Pastorino U, Bellomi M, Landoni C et al. Early lung-cancer detection with
spiral CT and positron emission tomography in heavy smokers: 2-year results.
Lancet 2003; 362: 593–7.
37. Antoch G, Stattaus J, Nemat AT et al. Non-small cell lung cancer: dual-modality
PET/CT in preoperative staging. Radiology 2003; 229: 526–33.
9781842143094-Ch08 8/8/07 12:02 PM Page 129
8
Tumors of the
gastro-intestinal tract
BACKGROUND
TECHNIQUE
Patient preparation
Adequate bowel preparation is required for accurate rectal cancer imaging:
24 h before the CTP examination a diet with low amounts of fiber is
observed and a cleansing enema is performed, combined with oral
laxative administration.
No oral contrast is administered to the patient. Once on the com-
puted tomography (CT) bed, a 24-Fr Foley catheter is positioned as low
as possible in the rectum, positioning the inflated balloon against the
inner margin of the anal sphincter, and a 2-liter water enema is admin-
istered for homogeneous, reproducible and persistent rectal wall disten-
tion and visualization. Non-contrast-enhanced CT of the pelvis (2.5 mm
slice thickness) is performed to locate the rectal cancer.
One milliliter of hyoscin butylbromide is administered immediately
before contrast medium injection and dynamic scanning, in order to
avoid movement artifacts due to bowel peristalsis.
the highest blood supply cannot be chosen prior to performing CTP, and
sampling is one of the main factors differentiating CTP from the patho-
logical evaluation of tumor vascularization, since the pathologist selects
the area with the highest vascularization to evaluate MVD.
First-pass imaging requires high-frequency sequences (at least one scan/s)
up to 50–60 s after initiation of contrast medium injection, when the
contrast medium is predominantly intravascular. Reliable assessment of
microvessel blood volume (BV) and accurate determination of functional
parameters such as blood flow (BF) and mean transit time (MTT) can be
obtained from the first-pass imaging. Delayed imaging of rectal cancer, to
evaluate contrast medium leakage into extravascular space, requires scan-
ning up to 2–10 min, at a lower frequency than first-pass imaging; these
series allow microvessel permeability to be computed with reliable and
reproducible accuracy.19–21 The correct scanning protocol, including
both first-pass and delayed imaging, may be designed as follows: 8–10 s
scanning delay from the start of contrast medium intravenous injection
(contrast medium is still in the pulmonary circuit, and the first 10 seconds
are not relevant for tumor perfusion), followed by continuous scanning
at high time resolution (1 s/scan) for 45–50 s and at low time resolu-
tion (10–15 s/scan) up to 3–4 minutes. For dose reduction, 100 kVp and
200–220 mAs should be used; however, for oncological applications the
dose might not be relevant, since many rectal cancer patients undergo
radiation therapy.
High spatial resolution with thin slices is not required for perfusion
imaging; on the contrary, thick slices (5–10 mm) are preferred since they
reduce image noise, which affects perfusion measurements. The acqui-
sition protocol for a 16-slice scanner should include two adjacent 10-mm
slices or four 5-mm slices. Spiral acquisitions increase volume coverage,
but will produce unacceptable time resolution for the perfusion imaging
of rectal cancer.
The image acquisition protocol is summarized in Table 8.1.
Slice thickness 4 × 5 mm or 2 × 10 mm
kVp 100
mAs 200–220
Dynamic scans Delay: 8–10 s
First pass: 45–50 s (one scan/s)
Delayed phase: 120–180 s (1 scan/10–15 s)
Contrast medium 300–370 mg/ml
40–70 ml
4-7 ml/s
18–20-gauge in the antecubital vein
a 219
5
200
180
160
Artery
Hounsfield units
140
120
100
5
80 5
60 2
2 2
40 2 Tumor tissue
21
0 10000 20000 30000 40000 49001 ms
Time (ms)
Figure 8.1 Graphs representing time (seconds) on the x-axis and computed
tomography (CT) attenuation–density (HU) on the y-axis. Time–density curves
can be obtained for different locations in the CT image: artery (a),
9781842143094-Ch08 8/8/07 12:02 PM Page 133
b 62
60
55 Rectal cancer 3 (1313.0)
52
3 (1313.0)
48 3 (1313.0)
44
Hounsfield units
40 3 (1313.0)
36
32
28
24 5 (1313.0)
5 (1313.0)
20 5 (1313.0)
16 5 (1313.0)
12 5 (1313.0)
8 Rectal wall
5
0 10000 20000 30000 40000 49001 ms
Time (ms)
c 65
64 3
Muscle 3 3
62 3
60 2
58
4
56
4
54 4
Hounsfield units
2
52 Lymph node
50
48 2 2
46
44
Tumor tissue
42
40 2
38
36
0 10000 20000 30000 40000 49001 ms
Time (ms)
Figure 8.1 cont’d rectal cancer and normal rectal wall (b), muscle and lymph
node (c). They reflect CT attenuation over time.
9781842143094-Ch08 8/8/07 12:02 PM Page 134
a b
c d
Figure 8.2 A 63-year-old patient with T3N1 rectal cancer. Regions of interest
(ROIs) are manually drawn on the original CT image (a): on rectal cancer
(lesion), normal rectal wall (normal wall), left external iliac artery (artery),
mesorectal pathologic lymph node (lymph node), and right gluteus muscle
(muscle). Functional maps of blood flow (b), blood volume (c), mean transit
time (d) and permeability surface area (e) are generated.
9781842143094-Ch08 8/8/07 12:02 PM Page 136
a b
c
Figure 8.3 A 56-year-old woman
with T3N1 rectal cancer, before
neoadjuvant chemoradiation ther-
apy. Original CT image (a) and
functional maps of blood flow (b)
and blood volume (c).
a b
c
Figure 8.4 The same woman as
Figure 8.3, after neoadjuvant
chemoradiation therapy. Original CT
image demonstrates tumor size reduc-
tion (a). Functional maps did not
show any significant difference in
blood flow (b) and blood volume (c)
between tumor and normal rectal
wall: pathologic stage revealed
absence of tumor (T0), after neo-
adjuvant treatment.
Figure 8.5 Conventional CT (a) and perfusion image (b) of a pancreatic islet
cell tumor. Note that perfusion in the tumor periphery is greater than in the
adjacent normal pancreas but perfusion is reduced in the center of the tumor.
(Reproduced with permission from reference 37)
9781842143094-Ch08 8/8/07 12:02 PM Page 141
Figure 8.6 cont’d hepatic arterial (b), and hepatic portal (c) perfusion images
in a patient with hepatic cirrhosis and a hepatocellular carcinoma in the right
lobe. The tumor demonstrates high arterial perfusion. Portal perfusion is
reduced in the remaining liver reflecting portal hypertension. (Reproduced with
permission from reference 37)
REFERENCES
22. Miles KA, Hayball MP, Dixon AK. Functional images of hepatic perfusion
obtained with dynamic CT. Radiology 1993; 188: 405–11.
23. Nabavi DG, Cenic A, Dool J et al. Quantitative assessment of cerebral hemo-
dynamics using CT: stability, accuracy, and precision studies in dogs. J Comput
Assist Tomogr 1999; 23: 506–15.
24. Harvey C, Morgan J, Blomley M et al. Tumor responses to radiation therapy:
use of dynamic contrast material-enhanced CT to monitor functional and
anatomical indices. Acad Radiol 2002; 9: S215–19.
25. Bellomi M, Petralia G, Sonzongni A, Zampino MG, Rocca A. CT perfusion for
the monitoring of neo-adjuvant chemoradiation therapy in rectal carcinoma –
initial experience. Radiology 2007; in press.
26. Rodel C, Martus P, Papadoupolos T et al. Prognostic significance of tumor
regression after preoperative chemoradiotherapy for rectal cancer. J Clin
Oncol 2005; 23: 8688–96.
27. Smith FM, Reynolds JV, Miller N et al. Pathological and molecular predictors
of the response of rectal cancer to neoadjuvant radiochemotherapy. Eur J Surg
Oncol 2006; 32: 55–64.
28. Leek RD, Landers RJ, Harris AL et al. Necrosis correlates with high vascular
density and focal macrophage infiltration in invasive carcinoma of the breast.
Br J Cancer 1999; 79: 991–5.
29. Wheeler RH, Ziessman HA, Medvec BR et al. Tumor blood flow and systemic
shunting in patients receiving intra-arterial chemotherapy for head and neck
cancer. Cancer Res 1986; 46: 4200–4.
30. Hlatky L, Hahnfeldt P, Folkman J. Clinical application of antiangiogenic therapy:
microvessel density, what it does and doesn’t tell us. J Natl Cancer Inst 2002;
94: 883–93.
31. Miles KA, Leggett DAC, Kelley BB et al. In-vivo assessment of neovascularisation
of liver metastases using perfusion CT. Br J Radiol 1998; 71: 276–81.
32. Mooteri S, Rubin D, Leurgans S et al. Tumor angiogenesis in primary and
metastatic colorectal cancers. Dis Colon Rectum 1996; 39: 1073–80.
33. Willett CG, Boucher Y, di Tomaso E et al. Direct evidence that the VEGF-
specific antibody bevacizumab has antivascular effects in human rectal cancer.
Nat Med 2004; 10: 145–7.
34. Blankenberg FG, Eckelman WC, Strauss HW et al. Role of radionuclide imaging
in trials of antiangiogenic therapy. Acad Radiol 2000; 7: 851–67.
35. Bacharach SL, Libutti SK, Carrasquillo JA. Measuring tumor blood flow with
H(2)(15)O: practical considerations. Nucl Med Biol 2000; 27: 671–6.
36. Miles KA, Hayball MP, Dixon AK. Measurement of human pancreatic perfusion
using dynamic computed tomography with perfusion imaging. Br J Radiol
1995; 68: 471–5.
37. Miles K, Blomley M. Applications of perfusion CT. In: Miles K, Blomley M,
Dawson P, eds. Functional Computed Tomography. Oxford: ISIS Medical
Media, 1997: 89–116.
38. Tsushima Y, Kusano S. Age-dependent decline in parenchymal perfusion in
the normal human pancreas: measurement by computed tomography. Pancreas
1998; 17: 148–52.
39. Abe H, Murakami T, Kubota M et al. Quantitative tissue blood flow evaluation
of pancreatic tumor: comparison between xenon CT technique and
perfusion CT technique based on deconvolution analysis. Radiat Med 2005;
23: 364–370.
9781842143094-Ch08 8/8/07 12:02 PM Page 145
9
Tumors of the urogenital tract
Elizabeth P Ives and Ethan J Halpern
For the 2007 calendar year, it is estimated that there will be 218 890 new
prostate cases and 27050 prostate cancer deaths in the United States.1 The
prostate is the leading site of new cancer diagnoses among US men and
the second leading cause of cancer deaths. For an American male, there
is a 1 in 6 lifetime probability of developing prostate cancer.
Traditional diagnostic imaging methods have a limited role in the
diagnosis, staging, and management of prostate cancer. The role of
imaging in the prostate is limited by poor discrimination of prostate
cancer from adjacent benign tissue, and by benign hypertrophic
changes within the prostate. Technological innovations in imaging have
the potential to improve the detection of prostate cancer with more
accurate biopsy techniques, better local staging, more precise focusing
of therapy, and closer post-treatment follow-up. Although computed
tomography (CT) currently has a small role in the diagnosis and stag-
ing of prostate disease,2 numerous reports indicate the frequent use
of CT by clinicians for evaluation of prostate cancer patients.3,4 Recent
improvements in multidetector CT technology permit quantitative
perfusion imaging of the prostate, and may presage a new role for CT
in the detection, staging, and management of prostate cancer.
Screening for prostate cancer is a controversial topic. Prostate spe-
cific antigen (PSA) serum levels and the digital rectal examination are
the current accepted standard of care for prostate cancer screening.
Transrectal ultrasound (TRUS) is not accepted as a screening method in
an unselected population because of the poor sensitivity and specificity
of sonography for prostate cancer. The accepted application for TRUS
is to guide a biopsy procedure of the prostate in a patient with an
abnormal serum PSA or digital rectal examination. Even if ultrasound-
guided biopsy can improve detection, it provides little information
about the extent of disease, because location and quantity of malig-
nancy on biopsy often do not correlate with actual cancer volume.5
9781842143094-Ch09 8/8/07 4:23 PM Page 148
a b
Figure 9.2 Patient with enlarged prostate secondary to benign prostatic hyper-
plasia. The pre-contrast image (a) demonstrates an enlarged gland. After con-
trast administration there is heterogeneous enhancement of the hypertrophied
transition zone that occupies most of the prostate (b)
9781842143094-Ch09 8/8/07 4:23 PM Page 150
Figure 9.3 Contour abnormality along the right side of the prostate corre-
sponds to an exophytic prostate cancer at the right mid-gland (arrow). Note that
this area of cancer does not enhance to a greater degree than other unaffected
parts of the prostate
9781842143094-Ch09 8/8/07 4:23 PM Page 151
a b
Figure 9.5 Normal CT and quantitative perfusion images. Images through the
mid-gland level of the prostate in a patient with a low-volume tumor in the apex
of the gland. Both the contrast-enhanced CT (a) and the quantitative perfusion
(b) demonstrate a normal, symmetric appearance
9781842143094-Ch09 8/8/07 4:23 PM Page 152
Kenneth A Miles
Renal cancer
CT perfusion studies of renal cancer were first reported in 1994.29 Renal
tumors are typically highly vascularized, but CT perfusion images
frequently show high perfusion peripherally with lower perfusion in a
central area of cystic change or necrosis29,30 (Figure 9.9). Metastatic
lesions from renal cancer are also frequently hypervascular (Figure 9.10).
Figure 9.10 Hepatic arterial (a) and portal (b) perfusion images of a hepatic
metastasis from renal cancer. Arterial perfusion values are very high, whereas
there is virtually no portal perfusion in the lesion. Note also abnormal arterial
and portal perfusion values in the adjacent liver tissue that is apparently normal
on conventional CT images. (Reproduced with permission from reference 31)
12
10
SPV 8
0
50 100 150 200
MVD (/mm2)
Cervical cancer
There has been a single but thorough study of CT perfusion in cancer
of the uterine cervix.35 The study group comprised 32 patients with
tumor stages ranging between T1b and T3b. Tumors demonstrated mod-
erately high vascularization. Based on deconvolution analysis, mean values
for perfusion, blood volume, and permeability were 95.5 ml/min/100 g,
9781842143094-Ch09 8/8/07 4:23 PM Page 159
Table 9.1 Peak enhancement values (35 s or 180 s) and corresponding perfusion/
cardiac output for various types of renal cancer (RCC) based on reference 34
*Assumes an iodine calibration factor of 21.32 HU/mg iodine as reported by Miles et al.33
REFERENCES
1. Jemal A, Siegel R, Ward E et al. Cancer statistics, 2007. CA Cancer J Clin 2007;
57: 43–66.
2. Levran A, Gonzalez, JA, Diokno AC et al. Are pelvic computed tomography,
bone scan, and pelvic lymphadenectomy necessary in the staging of prostatic
cancer? Br J Urol 1995; 75: 778–81.
3. Kindrick AV, Grossfeld GD, Stier DM et al. Use of imaging tests for staging
newly diagnosed prostate cancer: trends from the CAPSURE database. Urology
1998; 160: 2102–6.
4. Saigal CS, Pashos CL, Henning JM, Litwin MS. Variations in use of imaging in a
national sample of men with early-stage prostate cancer. Urology 2002; 59: 400–4.
5. Gardner TA, Lemer ML, Schlegel RS et al. Microfocal prostate cancer: biopsy
cancer volume does not predict actual tumour volume. Br J Urol 1998; 81:
839–43.
6. Dang CV, Semenza GL. Oncogenic alterations of metabolism. Trends Biochem
Sci 1999; 24: 68–72.
7. Bigler SA, Deering RE, Brawer MK. Comparison of microscopic vascularity in
benign and malignant prostate tissue. Human Pathol 1993; 24: 220–6.
9781842143094-Ch09 8/8/07 4:23 PM Page 160
27. Byar DP, Mostofi FK. Carcinoma of the prostate: prognostic evaluation of certain
pathologic features in 208 radical prostatectomies. Cancer 1972; 30: 5–13.
28. McNeal JE, Redwine EA, Freiha FS, Stamey TRA. Zonal distribution of prostatic
adenocarcinoma. Correlation with histologic pattern and direction of spread.
Am J Surg Pathol 1988; 12: 897–906.
29. Miles KA, Hayball MP, Dixon AK. Functional imaging of changes in human
intra-renal perfusion using quantitative dynamic computed tomography.
Invest Radiol 1994; 29: 911–14.
30. Wang JH, Min PQ, Wang PJ et al. Dynamic CT evaluation of tumor vascularity
in renal cell carcinoma. AJR Am J Roentgenol 2006; 186: 1423–30.
31. Miles K, Blomley M. Applications of perfusion CT. In: Miles K, Blomley M,
Dawson P, eds. Functional Computed Tomography. Oxford: ISIS Medical
Media, 1997: 89–116.
32. Miles KA, Griffiths MR, Fuentes MA. Standardized perfusion value: universal
CT contrast enhancement scale that correlates with FDG PET in lung nodules.
Radiology 2001; 220: 548–53.
33. Miles KA, Young H, Chica SL, Esser PD. Quantitative contrast-enhanced com-
puted tomography: is there a need for system calibration? Eur Radiol 2007; 17:
919–26.
34. Jinzaki M, Tanimoto A, Mukai M et al. Double-phase helical CT of small renal
parenchymal neoplasms: correlation with pathologic findings and tumor
angiogenesis. J Comput Assist Tomogr 2000; 24: 835–42.
35. Haider MA, Milosevic M, Fyles A et al. Assessment of the tumor microenviron-
ment in cervix cancer using dynamic contrast enhanced CT, interstitial fluid
pressure and oxygen measurements. Int J Radiat Oncol Biol Phys 2005; 62:
1100–7.
9781842143094-Ch09 8/8/07 4:23 PM Page 162
9781842143094-Ch10 8/8/07 12:05 PM Page 163
10
CT perfusion of lymph nodes
Kenneth A Miles
INTRODUCTION
high vascular contrast density can also produce beam hardening artifacts
within lymph nodes that lie close to major vessels. Although the appli-
cation of CT perfusion to the assessment of tumor within lymph nodes
is, as yet, underdeveloped, this chapter reviews the currently available
data supporting the use of this technique, both in the diagnosis of
lymph node metastases and in the assessment of lymphoma.
Table 10.1 Comparison of the diagnostic performances of size (single node ≥ 10 mm),
enhancement, and combined criteria in the diagnosis of nodal metastases
Enhancement Combined
Size criteria criteria criteria
Primary Sens. Spec. Sens. Spec. Sens. Spec.
tumor Authors (%) (%) (%) (%) (%) (%)
120
No metastasis
Metastasis
Enhancement (HU)
80
40
0
0 2 4 6 8 10
−40
Time (min)
a b
c d
40
30
20
HU
10
0
1b 2b 3b 4b
−10
Time (s)
T1 n1
40
Primary
Nodal mestastasis
30 Threshold
Size (mm) 20
10
0
0 Time
Figure 10.3 Theoretical growth curves for a primary tumor and nodal metastasis.
The double-headed green arrow demarcates a period of time in which enhance-
ment is readily measured in the primary tumor but the lymph node remains
normal by conventional size criteria
LYMPHOMA
a b
c d
SUMMARY
REFERENCES
11
CT perfusion of liver
metastases and early detection
of micrometastases
Charles A Cuenod, Laure Fournier, Daniel Balvay, and Kenneth A Miles
INTRODUCTION
The liver is the second site of metastatic disease,1 and the most
common site of metastasis of colorectal carcinoma.2 Eighty percent of
patients with colorectal carcinoma recurrence die within 3 years, pre-
dominantly from hepatic metastases.3
Despite the availability of various therapeutic options, hepatic metas-
tases remain difficult to eradicate, partly due to their late discovery.
Indeed, imaging techniques such as magnetic resonance imaging (MRI),
computed tomography (CT), and ultrasound (US) allow diagnosis of
most liver tumors over 1 cm (sensitivity of 55–90%),4 but, for smaller
lesions, sensitivity is much lower (under 50%), and microscopic lesions
remain occult. Liver metastases are present but not detected in up to
one-third of patients who undergo apparently curative excision of
primary colorectal carcinoma.5
Therefore, the identification of patients with micrometastatic disease
who are at risk of recurrence would have profound implications for
prognosis and treatment, since it would allow the implementation of
adjuvant chemotherapy at an earlier stage.6 On the other hand, exclu-
sion of micrometastastic disease could obviate potentially morbid
chemotherapy in patients without seeds of micrometastasis.
Hopes for the earlier detection of liver metastases have been raised
by functional imaging suggesting that occult metastases induce changes
in liver blood flow similar to those caused by overt metastases,7–11
such as a decrease in the portal perfusion of the liver and an increase
9781842143094-Ch11 8/8/07 4:25 PM Page 174
The liver has a unique perfusion system with a dual blood supply. More
than two-thirds of the blood supply comes from the low-pressure
portal vein, and the rest comes from the high-pressure hepatic artery
(Figure 11.1). The capillaries of the liver, called ‘sinusoid capillaries’,
therefore contain a mixture of portal and arterial blood.
The sinusoid capillary system is very dense, representing almost one-
third of the volume of the liver parenchyma (whereas in the brain, for
example, the blood within the capillaries contributes to only 4%
of the volume of tissue). The endothelial cells that line the sinusoids
are anatomically and biologically distinct from endothelial cells in
Hepatic veins
output C°(t)
Portal vein
input 2
Cp(t)
Figure 11.1 The liver is supplied more than two-thirds by the portal vein and
less than a third by the hepatic artery. A transverse slice through the abdomen
allows simultaneous measurement of the attenuation in regions of interst (ROIs)
drawn in the portal vein, in the aorta, and in the liver tissue
9781842143094-Ch11 8/8/07 4:25 PM Page 175
Introduction
The specific dual perfusion of the liver makes it more difficult to ana-
lyze with contrast-enhanced imaging than the perfusion of other tissues.
The enhancement curve of the liver, after the injection of a bolus of
contrast agent, is the combination of the enhancement due to the con-
trast agent flowing in the arterial blood and the contrast agent flowing
in the portal blood. Whereas the molecules of contrast agent arrive
9781842143094-Ch11 8/8/07 4:25 PM Page 176
quickly when they are delivered to the liver through the arterial route,
they are delayed and diluted by the splanchnic circulation when they
are delivered through the portal route.
Many strategies have been developed to take advantage of the portal
lag to separate and quantify the arterial and portal hepatic perfusions.
A comprehensive review of perfusion imaging of the liver has recently
been published by Pandharipande et al.16
in which CL, Ca, and Cp are the contrast agent concentrations measured
over time respectively within the liver, hepatic artery, and portal vein
derived from ROIs, and k1a, k1p, and k2 are the arterial and portal
venous inflow and liver outflow rate constants. By fitting measured
CL(t), the constants k1a, k1p, and k2 can be estimated, and can be used
to calculate hepatic arterial and portal venous perfusion, mean transit
time (MTT) of contrast agent through the liver, and contrast agent dis-
tribution volume within the liver (Vd). The distribution volume Vd of
contrast agent is used instead of the hepatic blood volume, because the
small-molecule contrast agent used in CT leaks freely and instanta-
neously across the sinusoid capillary wall, leading to a distribution
volume that associates the hepatic blood volume and part of the extra-
cellular Disse space.
9781842143094-Ch11 8/8/07 4:25 PM Page 178
H (t ) = ∫ h (τ ) d τ
t
0
9781842143094-Ch11 8/8/07 4:25 PM Page 179
h (t)
1.0
1.0
Residue function
R (t) R(t)=1-H(t)
Figure 11.2 Relations between the residue function R(t), the cumulative
frequency function H(t), and the probability density function h(t). The contrast
agent that progressively leaves the tissue accumulates outside the tissue. At the
venous outlet the concentration of contrast agent rises progressively before
decreasing to zero. The initial value of R(0) is normalized to one, as well, there-
fore, as the final value of H(ⴥ) and the area under the curve h(t). (Adapted
from reference 31)
R(t) = 1 − H(t)
where R(t) is the fraction of the impulsive input remaining within its
distribution volume Vd in the tissue at time t.31
The concentration–time curve of a tracer remaining in its volume of
distribution within the tissue Cd(t) can be predicted for any type of
input function, Ci(t), as the convolution of Ci(t) by R(t):
Ct(t) = Cd(t)⋅Vdt
9781842143094-Ch11 8/8/07 4:25 PM Page 180
Ci ( nT ) ⊗ R t ( nT ) = T ∑ kk == 0n−1 Ci ( nT − kT ) ⋅ R t ( kT )
⎡ ⎛ kT ⎞ c ⎤
C∗t ( nT ) = T ∑ kk==0n−1 ⎡⎣Ci ( nT − kT )⎤⎦ ⋅ a ⋅ exp ⎢− ⎜ ⎟ ⎥
⎢⎣ ⎝ b ⎠ ⎥⎦
The program yields the value of the three unknown factors a, b, and c,
allowing the estimation of Rt(t).
Since Rt(t) = Vd ⋅ R(t) and R(0) = 1, then Vd = Rt(0) and R(t) =
Rt(t)/Rt(0).
When R(t) has been worked out, h(t) can be obtained as its negative
derivative
dR ( t )
h(t) = −
dt
∫ tC ( t ) dt
∞
0
MTT = 0
∫ C ( t ) dt
∞
0 0
∫ th ( t ) dt
∞
MTT = 0
∫ h ( t ) dt
∞
∫ h ( t ) dt = 1 ∫ th ( t ) dt.
∞ ∞
and even more simply, knowing that, as MTT =
0 0
In the liver, arterial and portal blood are mixed in the sinusoidal capil-
laries, and the tissue concentration–time curve in the liver (referred to
as CL) can be expressed as:
C t ( t ) = ⎡⎣α Ca ( t ) + (1 − α ) Cp ( t ) ⎤⎦ ⊗ R t ( t )
⎡ ⎛ kT ⎞ ⎤
( )
c *t ( nT ) = T ∑ kk == 0n=1 αC a ( nT − kT ) + (1 − α ) Cp ( nT − kT ) ⋅ a ⋅ exp ⎢− ⎜ ⎟ c ⎥
⎣ ⎝ b⎠ ⎦
The program yields the value of the four unknown factors α, a, b, and
c, allowing the estimation of Rt(t) and α = HPI. Rt(t) allows the
calculation of MTT, Vdt, and FT, and HPI allows the calculation of
Fa = HPI × FT, and Fp = (1 − HPI) × FT.
80
60
∆HU
40
20
0
0 20 40 60 80 100
Time (s)
TABLE 11.1 The six main liver perfusion parameters that can be extracted with
functional computed tomography (CT)
20
10
0
0 10 20 30 40
−10
time (s)
Iiver
Figure 11.4 By analysis of the liver time–density curve (top left), it is possible
to derive parametric images displaying a range of parameters reflecting the
hepatic vasculature. (HPI; hepatic perfusion index)
a b
c d
Figure 11.5 Conventional portal phase computed tomography (CT) (a) and CT
perfusion images of a hepatic metastasis from colorectal cancer. The metastasis
exhibits increased arterial perfusion (b) and hepatic perfusion index (c) but low
levels of portal perfusion (d). Note that the area of increased arterial perfusion
and hepatic perfusion index extends beyond the margins of the metastasis
9781842143094-Ch11 8/8/07 4:25 PM Page 185
TABLE 11.2 Perfusion parameter values (median [range]) in rats with micrometastases
and macrometastases compared to control rats
Micrometastases
Heamodynamic Normal livers normal appearance
parameters control rats of liver Macrometastases
(Mann–Whitney U test, *p < 0.05, **p < 0.015) (Data from references 27 and 28)
Micrometastases
Modifications of liver perfusion can be found not only in patients with
overt liver metastases, but also in patients bearing occult microscopic
disease who reveal liver metastases on follow-up.
In a cohort of 80 patients with colon cancer and without overt liver
metastases, Platt et al.11 showed, on classic liver CT examination, that
liver enhancement 40 seconds post-injection was higher in 22 patients
who revealed liver metastases in the following 18-month period. With
a threshold of 0.4 for the ratio of the difference between the liver den-
sity before injection and that at 40 s over the maximum liver enhance-
ment, the test had a sensitivity of 75% and a specificity of 96%.
9781842143094-Ch11 8/8/07 4:25 PM Page 186
a 40 b 40 c 40
Enhancement (HU)
Enhancement (HU)
Enhancement (HU)
30 30 30
20 20 20
10 10 10
0 0 0
0 10 20 30 40 0 10 20 30 40 0 10 20 30 40
Time (s) Time (s) Time (s)
Figure 11.6 Three cases illustrating the relationship between hepatic enhance-
ment and survival. Conventional CT images are displayed above with corre-
sponding time–density curves (TDCs) below. The arrows indicate the separation
of arterial and portal phases of enhancement as identified by the time of peak
splenic enhancement. The first case (a) survived for more than 46 months fol-
lowing CT, which demonstrated no visible metastases on conventional images.
The hepatic TDC is normal with greater enhancement in the portal phase. The
second case (b) also had no visible metastases on CT but survived for a shorter
time (31 months). The hepatic TDC shows increased enhancement in the
arterial phase implying possible micrometastatic disease. The third case (c) had
visible metastases and survived only 9 months following CT. Arterial phase
enhancement is considerably raised
a 0.1 b 0.1
0.8 0.8
Survival
Survival
0.6 0.6
0.4 0.4
HPI < 0.35 no Mets Dukes A
0.2 HPI < 0.35 + Mets 0.2 Dukes B
HPI ≥ 0.35 no Mets Dukes C
HPI ≥ 0.35 + Mets Dukes D
0 0
0 12 24 36 0 12 24 36
Time (months) Time (months)
PHYSIOPATHOLOGY OF MICROCIRCULATION
ALTERATIONS IN METASTATIC LIVER
All the imaging techniques have shown similar flow alterations in liver
bearing visible metastases. The main finding is an increase in HPI (or
DPI with Doppler), which appears to be mostly due to a decrease in
portal perfusion, and to a lesser extent to an increase in arterial perfu-
sion. Such a relative increase in arterial supply in the environment of
liver metastases had already been described in 1954.50 An increase in
MTT and a decrease in distribution volume are also found.
Surprisingly, significant intrahepatic flow alterations also occur prior
to the growth of visible metastases. These hemodynamic changes are
similar but more discrete than those found in the presence of large
metastases.
The precise cause of these changes, however, remains only partly
explained. Both biomechanical and molecular mechanisms appear to
be implicated.
Humoral mediators have been implicated.51 The presence of circu-
lating vasoactive agents has been suggested by changes induced in
splanchnic hemodynamics in rats by injecting plasma samples from
patients or rats bearing tumors,52,53 but the nature of these mediators is
currently not known.
9781842143094-Ch11 8/8/07 4:25 PM Page 190
In vivo videomicroscopy
Direct observation using in vivo videomicroscopic techniques has made
it possible to study in live animals the flow alterations in hepatic metas-
tases and in the liver during their formation.54–57
Large tumor nodules contain irregularly dilated, tortuous, neoangio-
genic vascular networks. Blood flow velocity and direction vary in this
network, accounting for the increase of the mean transit time. The nod-
ules induce direct mechanical compression on the surrounding sinu-
soids, which are narrowed, and resist portal low-pressure flow. The
portal venules and the sinusoids are obliterated abruptly at the borders
of the growing metastatic foci,55 and most flow is stopped, accounting
for the drop in portal flow within the tumor and for the portal hyper-
tension in tumor-bearing liver. The drop in portal flow is compensated
by arterial flow58 via the arterial buffer effect, accounting for the
increase in HPI as well as the peritumoral rim and the transient segmen-
tal enhancements that can be observed during the arterial phase on CT
and MRI.59
Moreover, although metastases smaller than 200 µm receive their
blood supply via the sinusoids, without hepatic arterial neovasculariza-
tion,60 when tumors progress to a point beyond which angiogenesis is
necessary to sustain growth, neovascularization occurs,61 assisted by
vascular endothelial growth factor expression.62
Since significant intrahepatic flow alterations occur in livers before
metastases become visible, mechanisms other than extrinsic compres-
sion of sinusoids must be implicated. Intravital microscopy during
the formation of hepatic metastases showed that several intrahepatic
microhemodynamic events occur prior to the establishment of visible
metastases:57
• blocked metastatic cells within the sinusoids56
• increased post-sinusoidal leukocyte rolling and adherence, and stasis
of leukocytes in the sinusoids
• reduced wall shear rate (velocity gradient perpendicular to the direc-
tion of fluid flow)
• and, finally, a reduction in blood flow in the hepatic sinusoids.
Metastatic cells coming through the portal route are arrested at
the inflow side of the microvascular bed due to size restriction.56
Microthrombi form in the portal vessels55,60 and increase resistance in
the low-pressure vascular bed, which accounts for the increased MTT
and the decreased portal flow. Kruskal et al.57 showed that when com-
pared with normal livers, a reduction in sinusoidal flow velocity and
9781842143094-Ch11 8/8/07 4:25 PM Page 191
flow rate began as early as day 2 after the injection of tumor cells in the
portal system of mice that developed hepatic metastases, and they
noted a significant decrease in post-sinusoidal leukocyte velocity and
an increase in the number of stagnant adherent leukocytes, associated
with a decrease in wall shear rate. To resist substantial wall shear stress
exerted by blood flow, metastasizing cells have to form adhesive
contacts with endothelial cells. The reduced shear rate increases the
likelihood of interactions between leukocytes (and tumor cells) and
endothelial receptors, favoring formation of metastases. In return,
increased leukocyte rolling and adherence narrow the sinusoids and
portal venules and compromise flow.
Moreover, cell membrane molecules such as the carcinoembryonic
antigen, present in certain cancer cells including colon cancer cells,
bind to Kupffer cells by specific receptors, and activate them. Activated
Kupffer cells can release vasoactive cytokines that activate stellate cells,
which impair sinusoidal perfusion by modulating the sinusoidal caliber.
The increase in high-pressure arterial flow which partly compensates
for the portal flow decrease may be due to the buffer effect, or to
humoral factors also secreted by tumoral cells.
The low distribution volume of the contrast medium within the
tumor suggests that the capillary density is lower than in the normal
liver, and accounts for the fact that metastases from colon and breast
cancers usually appear to be less enhanced than normal liver after
contrast medium injection.55
CONCLUSION
REFERENCES
1. Paley MR, Ros P. Hepatic metastases. Radiol Clin North Am 1998; 36: 349–363.
2. Baker ME, Pelley R. Hepatic metastases: basic principles and implications for
radiologists. Radiology 1995; 197: 329–37.
3. Finlay IG, McArdle CS. Occult metastases in colorectal carcinoma. Br J Surg
1986; 73: 732–5.
4. Kinkel K, Lu Y, Both M, Warren RS, Thoeni RF. Detection of hepatic metas-
tases from cancers of the gastrointestinal tract by using noninvasive imaging
methods (US, CT, MR imaging, PET): a meta-analysis. Radiology 2002; 224:
748–56.
5. Greenway B. Hepatic metastases from colorectal cancer: resection or not.
Br J Surg 1988; 75: 513–19.
6. Moertel CG. Drug therapy: chemotherapy for colorectal cancer. N Engl J Med
1994; 330: 1136–42.
7. Cooke DA, Parkin A, Wiggins P, Robinson PJ, Giles GR. Hepatic perfusion
index and the evolution of liver metastases. Nucl Med Commun 1987; 8: 970–4.
8. Nott DM, Grime JS, Yates J et al. Changes in the hepatic perfusion index during
the growth and development of experimental hepatic micrometastases. Nucl
Med Commun 1987; 8: 995–1000.
9. Perkins AC, Whalley DR, Ballantyne KC, Hardcastle JD. Reliability of the
hepatic perfusion index for the detection of liver metastases. Nucl Med
Commun 1987; 8: 982–9.
10. Leen E, Goldberg JA, Robertson J et al. Image-directed Doppler ultrasonogra-
phy: a novel technique for the diagnosis of colorectal liver metastases. J Clin
Ultrasound 1993; 21: 221–30.
11. Platt JF, Francis IR, Ellis JH, Reige KA. Liver metastases: early detection based
on abnormal contrast material enhancement at dual-phased helical CT.
Radiology 1997; 205: 49–53.
12. Thimgan MS, Yee HF Jr. Quantitation of rat hepatic stellate cell contraction:
stellate cells, contribution to sinusoidal resistance. Am J Physiol Gastrointest
Liver Physiol 1999; 277: 137–43.
13. Shah V, Haddad FG, Garcia-Cardena G, et al. Liver sinusoidal endothelial cells
are responsible for nitric oxide modulation of resistance in the hepatic sinu-
soids. J Clin Invest 1997; 100: 2923–30.
14. Lautt WW, Legare DJ, Almeida MS. Adenosine as putative regulator of hepatic
arterial flow (the buffer reponse). Am J Physiol 1985; 248: 331–8.
15. Lautt WW. The 1995 Ciba-Geigy Award Lecture. Intrinsic regulation of hepatic
blood flow. Can J Physiol Pharmacol 1996; 74: 223–33.
16. Pandharipande PV, Krinsky GA, Rusinek H, Lee VS. Perfusion imaging of the
liver: current challenges and future goals. Radiology. 2005; 234: 661–73.
17. Miles KA, Hayball MP, Dixon AK. Functional images of hepatic perfusion
obtained with dynamic CT. Radiology 1993; 188: 405–11.
9781842143094-Ch11 8/8/07 4:25 PM Page 194
18. Dugdale PE, Miles KA. Hepatic metastases: the value of quantitative assess-
ment of contrast enhancement on computed tomography. Eur J Radiol 1999;
30: 206–13.
19. Blomley M, Coulden R, Dawson P. Liver perfusion studied with ultrafast CT.
J Comput Assist Tomogr 1995; 19: 424–33.
20. Tsushima Y, Blomley JK, Kusano S, Endo K. The portal component of hepatic
perfusion measured by dynamic CT: an indicator of hepatic parenchymal
damage. Dig Dis Sci 1999; 44: 1632–8.
21. Miles KA, McPherson SJ, Hayball MP. Transient splenic inhomogeneity on
contrast enhanced CT: mechanism and effect of liver disease. Radiology 1995;
194: 91–5.
22. Leggett DA, Kelley BB, Bunce IH, Miles KA. Colorectal cancer: diagnostic
potential of CT measurements of hepatic perfusion and implications for con-
trast enhancement protocols. Radiology 1997; 205: 716–20.
23. Miles KA, Leggett DA, Kelley BB et al. In vivo assessment of neovasculariza-
tion of liver metastases using perfusion CT. Br J Radiol 1998; 71: 276–81.
24. Halavaara JT, Hamberg LM, Leong FS et al. Functional CT with an experimen-
tal intravascular contrast agent in the assessment of liver vascular physiology.
Acad Radiol 1996; 3: 946–52.
25. Materne R, Van Beers BE, Smith AM et al. Non-invasive quantification of
liver perfusion with dynamic computed tomography and a dual-input one-
compartmental model. Clin Sci (Lond) 2000; 99: 517–25.
26. Van Beers BE, Leconte I, Materne R et al. Hepatic perfusion parameters
in chronic liver disease: dynamic CT measurements correlated with disease
severity. AJR Am J Roentgenol 2001; 176: 667–73.
27. Cuenod C, Leconte I, Siauve N et al. Early changes in liver perfusion caused
by occult metastases in rats: detection with quantitative CT. Radiology 2001;
218: 556–561.
28. Cuenod CA, Leconte I, Siauve N et al. Deconvolution technique for measuring
tissue perfusion by dynamic CT: application to normal and metastatic liver.
Acad Radiol 2002; 9: S205–11.
29. Fournier LS, Cuenod CA, de Bazelaire C et al. Early modifications of hepatic
perfusion measured by functional CT in a rat model of hepatocellular carci-
noma using a blood pool contrast agent. Eur Radiol 2004; 14: 2125–33.
30. Axel L. Tissue mean transit time from dynamic computed tomography by a
simple deconvolution technique. Invest Radiol 1983; 18: 94–19.
31. Zierler KL. Equation for measuring blood flow by external monitoring of
radioisotopes. Circ Res 1965; 16: 309–21.
32. Miles KA, Kelley BB. Altered perfusion adjacent to hepatic metastases. Clin
Radiol 1997; 52: 162–3.
33. Tsushima Y, Blomley MJ, Yokoyama H, Kusano S, Endo K. Does the presence
of distant and local malignancy alter parenchymal perfusion in apparently
disease-free areas of the liver? Dig Dis Sci 2001; 46: 2113–19.
34. Desch CE, Benson AB 3rd, Somerfield MR et al. American Society of Clinical
Oncology. Colorectal cancer surveillance: 2005 update of an American Society
of Clinical Oncology practice guideline. J Clin Oncol, 2005; 23: 8512–19.
35. Compton CC, Fielding LP, Burgart LJ et al. Prognostic factors in colorectal
cancer. College of American Pathologists Consensus Statement 1999. Arch
Pathol Lab Med 2000; 124: 979–94.
9781842143094-Ch11 8/8/07 4:25 PM Page 195
54. Ridge JA, Bading JR, Gelbard AS, Benua RS, Daly J. Perfusion of colorectal
hepatic metastases: relative distribution from the hepatic artery and portal vein.
Cancer 1987; 59: 1547–53.
55. Kan Z, Ivancev K, Lunderquist A et al. In vivo microscopy of hepatic tumors
in animal model: a dynamic investigation of blood supply to hepatic
metastases. Radiology 1993; 187: 621–6.
56. Morris VL, MacDonald IC, Koop S et al. Early interaction of cancer cells with
the microvasculature in mouse liver and muscle during hematogenous metas-
tasis: videomicroscopic analysis. Clin Exp Metastasis 1993; 11: 377–50.
57. Kruskal JB, Thomas P, Kane RA, Goldberg SN. Hepatic perfusion changes in
mice livers with developing colorectal cancer metastases. Radiology 2004; 231:
482–90.
58. Kopljar M, Brkljacic B, Doko M, Horzic M. Nature of Doppler perfusion index
changes in patients with colorectal cancer liver metastases. J Ultrasound Med
2004; 23: 1295–300.
59. Tian JL, Zhang JS. Hepatic perfusion disorders: etiopathogenesis and related
diseases. World J Gastroenterol 2006 May 28; 12: 3265–70.
60. Haugeberg G, Strohmeyer T, Lierse W. The vascularization of liver metastases.
J Cancer Res Clin Oncol 1988; 114: 415–19.
61. Terayama N, Terada T, Nakanuma Y. A morphometric and immunohisto-
chemical study on angiogenesis of human metastatic carcinomas of the liver.
Hepatology 1996; 24: 816–19.
62. Takeda A, Stoeltzing O, Ahmad SA et al. Role of angiogenesis in the develop-
ment and growth of liver metastasis. Ann Surg Oncol 2002; 9: 610–16.
63. Bader TR, Grabenwoger F, Prokesch RW, Krause W. Measurement of hepatic
perfusion with dynamic computed tomography: assessment of normal values
and comparison of two methods to compensate for motion artifacts. Invest
Radiol 2000; 35: 539–47.
64. Nakashige A, Horiguchi J, Tamura A et al. Quantitative measurement of
hepatic portal perfusion by multidetector row CT with compensation for
respiratory misregistration. Br J Radiol 2004; 77: 728–34.
65. Kapanen MK, Halavaara JT, Hakkinen AM. Assessment of vascular physiology
of tumorous livers: comparison of two different methods. Acad Radiol 2003;
10: 1021–9.
66. Kapanen MK, Halavaara JT, Hakkinen AM. Open four-compartment model in
the measurement of liver perfusion. Acad Radiol 2005; 12: 1542–50. Erratum
in: Acad Radiol 2006; 13: 270.
67. Jeffrey GM, Hickey BE, Hider P. Follow-up strategies for patients treated for
non-metastatic colorectal cancer (Cochrane Review). In: The Cochrane Library,
Issue 2, 2003. Oxford: Update Software, 2003.
9781842143094-Ch12 8/8/07 12:07 PM Page 197
12
Beyond RECIST: CT perfusion
in evaluating treatment
response and complications
Natalie Charnley and Kenneth A Miles
INTRODUCTION
a 140 b 14
120 12
Permeability (µl/min/ml)
Permeability (µl/min/ml)
100 10
80 8
60 6
40 4
20 2
0 0
Baseline With RMP-7 Baseline With RMP-7
Figure 12.1 Changes in the vascular permeability of gliomas (a) and normal
brain tissue (b) following RMP-7 assessed using computed tomography perfusion
(CT). (Data from reference 11)
9781842143094-Ch12 8/8/07 12:07 PM Page 200
1.5
ml/min/ml
0.5
Figure 12.3 CT perfusion images of non-small-cell lung cancer before (a) and
after (b) chemotherapy
the treatment response in lung cancer (Figure 12.3). Choi et al. meas-
ured peak tumor enhancement in residual masses in seven patients with
small-cell cancer following chemotherapy.20 Peak enhancement values
increased in four, decreased in one, and were unchanged in two. All
four patients demonstrating an increase in enhancement were reported
to exhibit improvement. As part of a larger study, Kiessling et al. pre-
sented a single case of non-small-cell lung cancer in which a partial
response on size criteria was associated with a reduction in perfusion.21
that agent are also assessed, as are early response data. MTD is relevant
to treatment with cytotoxic drugs, but not to many biological agents
such as antiangiogenic drugs and VDAs. Furthermore, there has been
no correlation between the activity of antiangiogenic agents and their
toxicity.22 Instead, surrogate endpoints must be investigated.
Angiogenesis is the formation of new tumor blood vessels, essential
for tumor growth from 1 mm3 in size.2 The trigger for angiogenesis
is additionally determined by hypoxia, acidosis, and hypoglycemia.23
The process of angiogenesis is tightly coordinated by a network of
cytokines. Vascular endothelial growth factor (VEGF) stimulates angio-
genesis and also has effects on vascular permeability. Fibroblast growth
factor (FGF) and platelet-derived growth factor (PDGF) also have a role
in angiogenesis, but are not vascular endothelium-specific. Integrins
such as αvβ3, which is expressed selectively on activated tumor vascu-
lature, potentiate angiogenesis and enable endothelial cell migration
towards the tumor.
Many of the steps in the angiogenic pathway have been targeted by
novel antiangiogenic drugs, and several of these are in clinical devel-
opment.22,24–27 Several of these trials compare CTP to other functional
imaging such as DCE-MRI, and also to histology, in proof of principle
studies. A summary of these trials is given in Table 12.1.
Willett’s group25 conducted a phase I trial of of bevacizumab, a VEGF-
specific antibody, in six patients with rectal adenocarcinoma. Patients
received a regimen of preoperative bevacizumab and chemoradiotherapy.
Perfusion was assessed initially 12 days following a single infusion of
the antiangiogenic agent. CTP revealed a significant decrease in tumor
perfusion of 40–44% and of blood volume of 16–39%, in four out of five
patients assessed. This was associated with a significant decrease in
tumor microvessel density (MVD). In addition, a fall in tumor intersti-
tial fluid pressure, and an increase in vessels positive for smooth muscle
actin, were observed. The authors suggest that this is supportive of
vascular normalization with bevacizumab.
Meijerink et al.24 assessed the perfusion of a range of advanced solid
tumors in a phase I trial of 13 patients using a combination of agents:
gefitinib, a tyrosine kinase inhibitor of epidermal growth factor receptor
1 (EGFR1), and the antiangiogenic agent AZD21271, a tyrosine kinase
inhibitor of VEGFR1–3. Both agents were given daily. CTP was per-
formed prior to starting treatment, and 4–6 weeks following treatment.
A longer scanning protocol was used for patients with liver metastases
compared to other sites, in order to determine the arterial and portal
fractions of tumor blood flow separately. In five out of six patients with
extrahepatic masses, there was an average decrease in perfusion of
9781842143094-Ch12
8/8/07
Yeung, 19943 10 Dexamethasone Brain Steroid 1 week ↓Transfer constant, ↓plasma volume
12:07 PM
Willett, 200425 5 Bevacizumab Rectum AA 12 days 4/5 ↓tumor BF, BV, MVD
McNeel, 200526 25 MEDI-522 Various AA 8 weeks ≠MTT with ≠doses of drug
Xiong, 200422 6 SU6668 Various AA 4 weeks, 5/6 ↓flow on CT
12 weeks
Falk, 199415 6 BW12C GI Binds to OHb 1h NS ↓tumor BF in 5/6 patients
Tozaki, 200418 19 Anthracycline + Breast Neoadjuvant NSt 82–100% accurate in detecting residual
taxane cytotoxic disease postoperatively
GI, gastrointestinal; AA, antiangiogenic agent; OHb, oxyhemoglobin; NSt, not stated: assessment made after 4–6 cycles of chemotherapy, likely to be 12–18 weeks;
BBB, blood–brain barrier; BF, blood flow; mets, metastases; BV, blood volume; MVD, microvessel density; MTT, mean transit time; NS, non-significant
Beyond RECIST: evaluating treatment response
203
9781842143094-Ch12 8/8/07 12:07 PM Page 204
model used are not always described. However, none of the above
studies in which multiple parameters were studied showed significant
changes in vascular permeability on treatment, even when significant
changes in other parameters had been observed.22,25,26 This finding is
interesting, considering the importance of VEGF not only in stimulating
angiogenesis but also in determining capillary leakiness. CT measure-
ments of permeability effectively describe the rate of transfer of contrast
material between intravascular and extravascular compartments. It is
feasible that in some circumstances, a decrease in transport of contrast
material due to a treatment-induced reduction in vascular permeability
may be offset by an increase in transport resulting from a concomitant
reduction in tumor interstitial fluid pressure.
When evaluating tumor response, the time scale for imaging is of
paramount importance. Frequently, changes are assessed at times when
we would expect to see a response on conventional cross-sectional
imaging, but this may not be the most appropriate time point for anti-
angiogenic drugs and VDAs. Response should not be assessed too
early, as during the first few days following an antiangiogenic drug, it
is thought that the vasculature undergoes some normalization,28 which
may lead to a reduction in interstitial fluid pressure and an initial pos-
sible increase in perfusion. However, assessing at late time points may
cause a change in tumor size and difficulty in co-registering the regions
of interest.
flow in 22 patients with carcinoma of the prostate, both 1–2 weeks and
6–12 weeks following completion of treatment.29 Increases in perfusion,
contrast agent clearance, and fractional vascular volume were observed
1–2 weeks following treatment, and these were maintained at the later
time point. Increases in contrast agent clearance and fractional vascular
volume, but not perfusion, were also seen at the same time points in
other tumor types.30
CTP has also been used to assess patients undergoing combined
modality treatment with chemoradiotherapy.31 Nine patients with rectal
carcinoma underwent a 6–8-week course of chemoradiation prior to
surgery. Patients were assessed following completion of treatment.
A reduction in BF and increase in MTT were observed, suggesting
damage to vasculature. BV and PS remained constant.
Response to other forms of radiation treatment can also be studied
using CTP. Bondestam et al.32 have investigated perfusion parameters
in six patients with meningiomas, undergoing brachytherapy treatment
by stereotactic implantation of iodine-125 seeds. A 41% reduction in
tumor blood flow was seen 3 months following the procedure using
CTP imaging.
OTHER THERAPIES
a b 50HU
0HU
d 50HU
c
0HU
results are in contrast with the studies above in that a high initial BF
and a short MTT were associated with poor response. The presence of
intratumoral arteriovenous shunts was proposed as a possible explana-
tion for these findings.
The use of imaging to identify patients unlikely to respond to cancer
therapy has the potential to save the morbidity and cost of futile treat-
ment. However, the diagnostic thresholds chosen must favor high
specificity over sensitivity in order to ensure that treatment is not with-
held from potential responders in error. This use of imaging is also
affected by the prevalence of non-responders in the treated population,
and is more useful when the proportion of patients failing to respond
9781842143094-Ch12 8/8/07 12:07 PM Page 209
CONCLUSIONS
4
Perfusion (ml/min/ml)
0
1 1 2 2 3 4 4 5 Norm
Patient number
a b
c d
Figure 12.6 Hepatic arterial (a) and portal (b) perfusion images in a patient
with graft-versus-host disease following bone-marrow transplantation. Hepatic
arterial perfusion is markedly reduced throughout the liver. The changes are
comparable to those seen in liver allograft rejection (c, d)
REFERENCES
19. Gandhi D, Chepeha DB, Miller T et al. Correlation between initial and early
follow-up CT perfusion parameters with endoscopic tumor response in
patients with advanced squamous cell carcinomas of the oropharynx treated
with organ-preservation therapy. Am J Neuroradiol 2006; 27: 101–6.
20. Choi J-B, Park C-K, Park DW et al. Does contrast enhancement on CT suggest
tumor response for chemotherapy in small cell carcinoma of the lung?
J Comput Assist Tomogr 2002; 26: 797–800.
21. Kiessling F, Boese J, Corvinus C et al. Perfusion CT in patients with advanced
bronchial carcinomas: a novel chance for characterisation and treatment
monitoring? Eur Radiol 2004; 14: 1226–33.
22. Xiong HQ, Herbst R, Faria SC et al. A phase I surrogate endpoint study of
SU6668 in patients with solid tumors. Invest New Drugs 2004; 22: 459–66.
23. Carmeliet P, Jain RK. Angiogenesis in cancer and other diseases. Nature 2000;
407: 249–57.
24. Meijerink MR, van Cruijsen H, Hoekman K et al. The use of perfusion CT for
the evaluation of therapy combining AZD2171 with gefitinib in cancer patients.
Eur Radiol 2006 Oct 27; [EPub ahead of print].
25. Willett CG, Boucher Y, di Tomaso E et al. Direct evidence that the VEGF-
specific antibody bevacizumab has antivascular effects in human rectal cancer:
Nat Med 2004; 10: 145–7.
26. McNeel DG, Eickhoff J, Lee FT et al. Phase I trial of a monoclonal antibody
specific for integrin (MEDI-522) in patients with advanced malignancies,
including an assessment on tumor perfusion. Clin Cancer Res 2005; 11: 7851–60.
27. Thomas JP, Arzoomanian RZ, Alberti D et al. Phase I pharmacokinetic and
pharmaocodynamic study of recombinant human endostatin in patients with
advanced solid tumors. J Clin Oncol 2003; 21: 223–31.
28. Winkler F, Kozin SV, Tong RT et al. Kinetics of vascular normalisation by
VEGFR2 blockade governs brain tumor response to radiation: role of oxygen-
tation, angiopoietin-1 and matrix metalloproteases. Cancer Cell 2004; 6: 553–63.
29. Harvey CJ, Blomley MJ, Dawson P, et al. Functional CT imaging of the acute
hyperaemic response to radiation therapy of the prostate gland: early experi-
ence. J Comput Assist Tomogr 2001; 25: 43–9.
30. Harvey C, Dooher A, Morgan J et al. Imaging of tumor therapy responses by
dynamic CT. Eur J Radiol 1999; 30: 221–6.
31. Sahani DV, Kalva SP, Hamberg LM et al. Assessing tumor perfusion and treat-
ment response in rectal cancer with multisection CT: initial observations.
Radiology 2005; 234: 785–92.
32. Bondestam S, Halavaara JT, Jaaskelainen JE et al. Perfusion CT of the brain in
the assessment of flow alterations during brachytherapy of meningioma. Acta
Radiol 1999; 40: 469–73.
33. Tsushima Y, Funabasama S, Aoki J, Sanada S, Endo K. Quantitative perfusion
map of malignant liver tumors created from dynamic computed tomography
data. Acad Radiol 2004; 11: 215–23.
34. Purdie TG, Sherar MD, Lee TY. The use of CT perfusion to monitor the effect
of hypocapnia during laser thermal therapy in a rabbit model. Int J
Hypertherm 2003; 19: 461–79.
35. Sommerfeld NWB, Miles KA, Dugdale P, Leggett DAC, Bunce IH. Colorectal
cancer: progressive disease is associated with altered liver perfusion on functional
CT. Abstract presented at 50th Annual Meeting of the Royal Australian and
New Zealand College of Radiologists, 21–25 October, 1999, Sydney: Abstr No 56.
9781842143094-Ch12 8/8/07 12:07 PM Page 214
36. Hermans R, Meijerink M, Van den Bogaert W et al. Tumor perfusion rate deter-
mined noninvasively by dynamic computed tomography predicts outcome in
head-and-neck cancer after radiotherapy. Int J Radiat Oncol Biol Phys 2003;
57: 1351–6.
37. Millar BM, Purdie TG, Yeung I et al. Assessing perfusion changes during
whole brain irradiation for patients with cerebral metastases. J Neurooncol
2005; 71: 281–6.
38. Miles KA; Hayball MP, Dixon AK. Functional imaging of changes in human
intra-renal perfusion using quantitative dynamic computed tomography. Invest
Radiol 1994; 29: 911–14.
39. Mulder WJ, Strijkers GJ, Habets JW et al. MR molecular imaging and fluores-
cence microscopy for identification of activated tumor endothelium using a
bimodal lipidic nanoparticle. FASEB J 2005; 19: 2008–10.
40. Wang ZQ, Li JS, Lu GM et al. Correlation of CT enhancement, tumor angio-
genesis and pathologic grading of pancreatic carcinoma. World J Gastroenterol
2003; 9: 2100–4.
9781842143094-Ch13 8/8/07 12:08 PM Page 215
13
Perfusion CT–PET:
opportunities for combined
assessment of tumor
vascularity and metabolism
Kenneth A Miles
a b
c d
Oncogene
Tissue mutations
hypoxia e.g. p53
HIF
Upregulated
Nitric oxide VEGF Glut-1 MDR p-Gp
molecules
Vascular
Biological Increased Multidrug
dilatation and Angiogenesis
effects glucose metabolism resistance
permeability
Table 13.1 Summary of imaging studies comparing tumor vascularity and metabolism
CBV, cerebral blood volume; CCC, cholangiocarcinoma; DC-MRI, dynamic contrast-enhanced magnetic
resonance imaging; FDG-PET, fluorodeoxyglucose-positron emission tomography; HCC, hepatocellular
carcinoma; H215O, oxygen-15-labeled water; mets, metastases; NSLLC, non-small-cell lung cancer;
Perf CT, computed tomography perfusion.
0
SPV = 5.5 20 SUV = 2.8
a b 0 c
Figure 13.4 Conventional CT (a), CT perfusion (b), and FDG-PET (c) images
of malignant (upper row) and inflammatory (lower row) lung lesions. In the
malignant lesion, glucose metabolism (expressed as the standardized uptake
value (SUV)), exceeds perfusion (expressed as the standardized perfusion value
(SPV)), whereas the flow–metabolic relationship is reversed in the inflammatory
lesion
9781842143094-Ch13 8/8/07 12:08 PM Page 222
Before
therapy
After
therapy
CT Perfusion CT FDG-PET
Table 13.2 Summary of studies assessing changes in both tumor perfusion and glucose metabolism following therapy
Kurdziel et al.36 Androgen-independent Thalidomide H215O-PET Positive correlation between prostate specific antigen (PSA)
prostate cancer FDG-PET response and change in glucose metabolism. Negative
correlation between PSA response and change
in perfusion
Willett et al.37 Rectal Bevacizumab Perf CT Significant reduction in perfusion. No change in FDG
cancer FDG-PET uptake
9781842143094-Ch13 8/8/07 12:08 PM Page 225
FDG-PET has been used to evaluate the kinetics of liver glucose metab-
olism using a three-compartment model with four rate constants (k1–4)
that define the rates of transfer between plasma FDG, tissue FDG, and
tissue FDG-6-phosphate (Figure 13.6).39 CT measurements of hepatic
perfusion can feed into this kinetic analysis by providing an estimate
of the rate of passage between intravascular and intracellular non-
phosphorylated compartments, k1.40 The permeability of the liver sinu-
soids to glucose is sufficiently high for this rate constant approximate
perfusion, as confirmed by the studies of Munk et al.41 Hence, the ratio
of hepatic FDG uptake to hepatic perfusion provides an index of the
hepatic phosphorylation fraction (HPFI: see Box 13.1).
Applying the above methodology to patients with colorectal cancer
it has been possible to demonstrate reduced hepatic phosphorylation of
k1 k3
Plasma Tissue Tissue
FDG FDG FDG-6-PO4−
k2 k4
Using the three-compartmental model described by Choi et al. (Figure 13.6),39 the fraction
of intracellular glucose that undergoes phosphorylation (i.e. the phosphorylation fraction,
PF) is given by:
PF = k3/(k2 + k3) (1)
The net influx constant, Ki is given by:
Ki = (k1k3)/(k2 + k3) = k1 × PF (2)
If the standardised uptake value (SUV) of FDG in the liver is used as a surrogate for the
net influx constant (Ki) and CT perfusion measurements of combined arterial and portal
perfusion as a surrogate for k1, then an index of hepatic phosphorylation (HPFI) can be
obtained from:
HPFI = SUV / Hepatic perfusion (3)
SUMMARY
REFERENCES
14. Meert AP, Paesmans M, Martin B et al. The role of microvessel density on the
survival of patients with lung cancer: a systematic review of the literature with
meta-analysis. Br J Cancer 2002; 87: 694–701.
15. Tanaka F, Yanagihara K, Otake Y, et al. Prognostic factors in resected patho-
logic (p-) stage IIIA-N2, non-small-cell lung cancer. Ann Surg Oncol 2004; 11:
612–18.
16. Giatromanolaki A, Koukourakis MI, Sivridis E et al. Relation of hypoxia
inducible factor 1 alpha and 2 alpha in operable non-small cell lung cancer
to angiogenic/molecular profile of tumors and survival. Br J Cancer 2001;
85: 881–90.
17. Aronen HJ, Pardo FS, Kennedy DN et al. High microvascular blood volume is
associated with high glucose uptake and tumor angiogenesis in human
gliomas. Clin Cancer Res 2000; 6: 2189–200.
18. Mankoff DA, Dunnwald LK, Gralow JR et al. Blood flow and metabolism in
locally advanced breast cancer: relationship to response to therapy. J Nucl Med
2002; 43: 500–9.
19. Semple SI, Gilbert FJ, Redpath TW et al. The relationship between vascular
and metabolic characteristics of primary breast tumors. Eur Radiol 2004; 14:
2038–45.
20. Fukuda K, Taniguchi H, Koh T, Kunishima S, Yamagishi H. Relationships
between oxygen and glucose metabolism in human liver tumors: positron
emission tomography using (15)O and (18)F-deoxyglucose. Nucl Med
Commun 2004; 25: 577–83.
21. van Laarhoven HW, de Geus-Oei LF, Wiering B et al. Gadopentetate dimeglu-
mine and FDG uptake in liver metastases of colorectal carcinoma as deter-
mined with MR imaging and PET. Radiology 2005; 237: 181–8.
22. Tateishi U, Nishihara H, Tsukamoto E et al. Lung tumors evaluated with FDG-
PET and dynamic CT: the relationship between vascular density and glucose
metabolism. J Comput Assist Tomogr 2002; 26: 185–90.
23. Hoekstra CJ, Stroobants SG, Hoekstra OS et al. Measurement of perfusion in
stage IIIA-N2 non-small cell lung cancer using H(2)(15)O and positron emis-
sion tomography. Clin Cancer Res 2002; 8: 2109–15.
24. Miles KA, Griffiths MR, Keith CJ. Blood flow-metabolic relationships are
dependent on tumor size in non-small cell lung cancer: a study using quanti-
tative contrast-enhanced computer tomography and positron emission tomog-
raphy. Eur J Nucl Med Mol Imaging 2006; 33: 22–8.
25. Veronesi G, Landoni C, Pelosi G et al. Fluoro-deoxi-glucose uptake and angio-
genesis are independent biological features in lung metastases. Br J Cancer
2002; 86: 1391–5.
26. Swensen SJ, Viggiano RW, Midthun DE et al. Lung nodule enhancement at CT:
multicenter study. Radiology 2000; 214: 73–80.
27. Yamashita K, Matsunobe S, Tsuda T et al. Intratumoral necrosis of lung carci-
noma: a potential diagnostic pitfall in incremental dynamic computed tomogra-
phy analysis of solitary pulmonary nodules? J Thorac Imaging 1997; 12: 181–7.
28. Miles KA, Griffiths MR, Fuentes MA. Standardized perfusion value: universal
CT contrast enhancement scale that correlates with FDG PET in lung nodules.
Radiology 2001; 220: 548–53.
29. Hermans R, Meijerink M, Van den Bogaert W et al. Tumor perfusion rate deter-
mined noninvasively by dynamic computed tomography predicts outcome in
9781842143094-Ch13 8/8/07 12:08 PM Page 229
head-and-neck cancer after radiotherapy. Int J Radiat Oncol Biol Phys 2003;
57: 1351–6.
30. Allal AS, Slosman DO, Kebdani T et al. Prediction of outcome in head-and-
neck cancer patients using the standardized uptake value of 2-[18F]fluoro-2-
deoxy-D-glucose. Int J Radiat Oncol Biol Phys 2004; 5: 1295–300.
31. Devries AF, Griebel J, Kremser C et al. Tumor microcirculation evaluated by
dynamic magnetic resonance imaging predicts therapy outcome for primary
rectal carcinoma. Cancer Res 2001; 61: 2513–16.
32. Sahani DV, Kalva SP, Hamberg LM et al. Assessing tumor perfusion and treat-
ment response in rectal cancer with multisection CT: initial observations.
Radiology 2005; 234: 785–92.
33. Oku S, Nakagawa K, Momose T et al. FDG-PET after radiotherapy is a good
prognostic indicator of rectal cancer. Ann Nucl Med 2002; 16: 409–16.
34. Herbst RS, Mullani NA, Davis DW et al. Development of biologic markers of
response and assessment of antiangiogenic activity in a clinical trial of human
recombinant endostatin. J Clin Oncol 2002; 20: 3804–14.
35. Mankoff DA, Dunnwald LK, Gralow JR et al. Changes in blood flow and
metabolism in locally advanced breast cancer treated with neoadjuvant
chemotherapy. J Nucl Med 2003; 44: 1806–14.
36. Kurdziel KA, Figg WD, Carrasquillo JA et al. Using positron emission tomog-
raphy 2-deoxy-2-[18F]fluoro-D-glucose, 11CO, and 15O-water for monitoring
androgen independent prostate cancer. Mol Imaging Biol 2003; 5: 86–93.
37. Willett CG, Boucher Y, di Tomaso E et al. Direct evidence that the VEGF-
specific antibody bevacizumab has antivascular effects in human rectal cancer.
Nat Med 2004; 10: 145–7.
38. Miles KA. Computed tomography measurements of perfusion in cancer therapy.
In: Shields AF, Price P, eds. In vivo Imaging of Cancer Therapy Totowa, NJ:
Humana Press, 2007: in press.
39. Choi Y, Hawkins RA, Huang SC et al. Evaluation of the effect of glucose inges-
tion and kinetic model configurations of FDG in the normal liver. J Nucl Med
1994; 35: 818–23.
40. Miles KA, Comber L, Keith CJ et al. Combining perfusion CT and FDG-PET
data demonstrates reduced hepatic phosphorylation of glucose in patients with
advanced colorectal cancer and poor survival. Eur Radiol 2003; 13(Suppl 1): 293.
41. Munk OL, Bass L, Roelsgaard K et al. Liver kinetics of glucose analogs
measured in pigs by PET: importance of dual-input blood sampling. J Nucl
Med 2001; 42: 795–801.
42. Tayek JA. A review of cancer cachexia and abnormal glucose metabolism in
humans with cancer. J Am Coll Nutr 1992; 11: 445–56.
43. Capra S, Ferguson M, Ried K. Cancer: impact of nutrition intervention out-
come—nutrition issues for patients. Nutrition 2001; 17: 769–72.
44. Laviano A, Meguid MM, Inui A, Muscaritoli M, Rossi-Fanelli F. Therapy insight:
cancer anorexia-cachexia syndrome—when all you can eat is yourself. Nat
Clin Pract Oncol 2005; 2: 158–65.
45. Warburg O, Wind F, Negelin E. The metabolism of tumors in the body. J Gen
Physiol 1927; 8: 519–30.
46. Folkman J. Tumor angiogenesis: therapeutic implications. N Engl J Med 1971;
285: 1182–6.
9781842143094-Ch13 8/8/07 12:08 PM Page 230
9781842143094-Idx 8/8/07 5:37 PM Page 231
Index
232 Index
Index 233
234 Index
Index 235
236 Index
Index 237
238 Index
Index 239
240 Index
Index 241