University of Southern California Dissertations and Theses

Noninvasive in vivo MRI measurements of tumor vascularization and relative tumor osmotic pressure

(USC Thesis Other)

Noninvasive in vivo MRI measurements of tumor vascularization and relative tumor osmotic pressure

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content NONINVASIVE IN VIVO MRI MEASUREMENTS OF TUMOR
VASCULARIZATION AND RELATIVE TUMOR OSMOTIC PRESSURE
by
Hyun Kwon Kim
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(PHARMACEUTICAL SCIENCES)
May 2001
Copyright 2001 Hyun Kwon Kim
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 3110952
Copyright 2004 by
Kim, Hyun Kwon
All rights reserved.
INFORMATION TO USERS
The quality of this reproduction is dependent upon the quality of the copy
submitted. Broken or indistinct print, colored or poor quality illustrations and
photographs, print bleed-through, substandard margins, and improper
alignment can adversely affect reproduction.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if unauthorized
copyright material had to be removed, a note will indicate the deletion.
®
UMI
UMI Microform 3110952
Copyright 2004 by ProQuest Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
ProQuest Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UNIVERSITY OF SOUTHERN CALIFORNIA '
TH E GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES. CALIFORNIA 90007
This dissertation, written by
B i u ^ . .., ]< Q. V . O . .......................
under the direction of h i s Dissertation
Committee, and approved by all its members,
has been presented to and accepted by The
Graduate School, in partial fulfillment of re
quirements for the degree of
DOCTOR OF PHILOSOPHY
Dean of Graduate Studies
D a te ....
PTFFFRTATTONJ m M M T T T F F y
Chairperson
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ACKNOWLEDGMENTS
I am sincerely grateful to the many people who gave me advice, assistance
and encouragement during the course of this work.
In particular I am indebted to my advisor, Professor W. Wolf for his
positive direction and many useful and encouraging discussions for this
manuscript. I would also like to thank Victor Waluch M.D. Ph.D. for the use
of their MRI facility.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
i i i
TABLE OF CONTENTS
0.1 TITLE i
0.2 ACKNOW LEDGM ENTS ii
0.3 TABLE OF CO N TEN TS iii
0.4 LIST OF FIGURES vi
0.5 LIST OFTABLES viii
0.6 ABSTRACT ix
1.0 SECTION 1 (GENERAL INTRODUCTION) 1
1.1 TUMOR GROWTH AND ANGIOGENESIS 1
1.1.1 PREVASCUALR PHASE 1
1.1.2 ONSETOF ANGIOGENIC ACTIVITY 2
1.1.3 ANGIOGENIC DEPENDENCEOFTUMORGROUTH 3
1.1.4 TEMPORAL ONSET OF ANGIOGENESIS 3
1.1.5 TUMOR STRUCTURE AND ITS ENVIRONMENT 5
1.2 BLOOD FLOW AND OXYGENATION 8
1.3 PAST AND PRESENT USE OF 5FU 10
1.3.1 5FU 10
1.3.2 FLUORINE: 5FU STRUCTURE AND PROPERTIES 10
1.3.2.1 FLUORINE 10
1.3.2.2 5-FU 11
1.3.2.3 BIOCHEMISTRY OF 5FU 12
1.3.3 MECHANISM OF ACTION OF 5FU 17
AS AN ANTITUMOR AGENT
1.3.3.1 INHIBITION OF THYMIDYLATE SYNTHETASE 17
1.3.3.2 INCORPORATION OFFDUTP INTO DNA 19
1.3.3.3 INCORPORATION OF FDUTP INTO RNA 21
1.3.3.4 ALTERATION OF CELLULAR MEMBRANES
AND DECREASED GLYCOPROTEIN SYNTHEIS 22
1.4 PHARMACOKINETICS (PK) OF5FU 23
1.4.1 INACTIVATION OF 5FU 28
1.4.2 DISTRIBUTION FOLLOWING iv INJECTION 29
1.4.3 RESISTANCE T05FU 30
1.5 CANCER M ODELS 33
1.5.1 K EY FACTORS 35
1.5.2 SPATIAL VARIATIONS 35
1.6 MAGNETIC RESONANCE IMAGING (MRI) 59
1.6.1 GENERAL THEORYOFMRI 59
1.6.2 Ti RELAXATION 63
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
iv
1.6.3 T2 RELAXATION 64
1.6.4 FORMING IMAGES 65
2.0 SPECIFIC AIMS 69
3.0 SURFACE COIL 70
3.1 SURFACE COIL DESIGN and PRACTICE 70
3.1.1 INTRODUCTION 70
3.1.2 SHIELDING 73
3.1.3 COUPLING 74
3.1.4 T .-C RATIO 79
3.1.5 SENSITIVITY VS. HOM OGENEITY 80
3.2 SURFACE COIL EXPERIMENT 83
3.2.1 COIL REQUIREMENTS 83
3.2.2 CIRCUIT DESCRIPTION 83
3.2.3 COIL TESTING (BENCH WORK) 86
3.2.4 COIL TESTING (in vivo) 86
3.2.5 COIL MAPPING 88
3.2.6 VOI DETERMINATION WITH
RESPECT TOTHE COIL 89
3.3 IN VIVO STUDIES 91
3.4 RESULTS AND DISCUSSION 92
4.0 MAGNETIC RESONACE SPECTROSCOPY (MRS) 94
4.1 MAGNETIC RESONACE SPECTROSCOPY 94
4.1.1 NUCLEI 94
4.1.2 MR SIGNALS 95
4.1.3 CHEMICAL SHIFT 95
4.1.4 SPIN-SPIN COUPLING 97
4.1.5 OTHER NUCLEI 97
4.2 5-FLUOROURACIL (5FU) 99
4.2.1 5-FLUOROURACIL 99
4.2.2 TRAPPING 100
4.3 M ETHOD 104
4.4 RESULTS 1Q 4
4.5 DISCUSSION 109
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
V
4.5.1 5FU METABOLITES 110
4.5.2 5FU LIMITATIONS 112
5.0 DYNAMIC ENHANCED MRI (DEMRI) 113
5.1 DYNAMIC ENHANCED MRI (CONTRAST AGENT) 113
5.1.1 Theory of DEMRI 113
5.1.2 Gadolinium Chelate 118
5.1.3 Gd-DTPA 119
5.1.4 DEMRI Profile 121
5.1.4.1 Pre-Contrast 122
5.1.4.2 ICAR 123
5.1.4.3 DCAR 124
5.1.4.4 Mathematical Model 125
5.2 M ETH O D 131
5.3 RESULTS 134
5.4 DISCUSSION 137
6.0 COMBINED DISCUSSION 139
6.1 5FU FIELD 146
6.1.1 5FU METABOLITES 148
6.2 DEMRI FIELD (Gd-DTPA) 149
6.2.1 ICAR 149
6.2.2 DCAR 150
6.2.3 INTERSECTING FIELD 5FU/DEMRI (ICAR) 150
6.2.4 INTERSECTING FIELD 5FU/DEMRI (DCAR) 151
7.0 CONCLUSIONS 153
8.0 FUTURE WORKS 157
9.0 REFERENCES 159
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF FIGURES
Figure List of Figures Page
1.1 Structure of 5-fluorouracil (5FU) and uracil. 12
1.2 Anabolites of 5FU. 14
1.3 Ternary complex of 5,10-m ethylene-tetrahydrofolate. 18
1.4 Effect of 5FU on DNA synthesis. 20
1.5 Curves for 5FU andF-DHU post iv bolus injection. 24
1.6 Rotation of the bulk mag. by 90 deg. away from B0. 62
1.7 Precessing proton. 62
1.8 Tj relaxation. 63
1.9 T2 relaxation, loss of processional coherence. 65
3.1 Double tuned STL coil and 500 coaxial line. 71
3.2 STL coil coupling profile. 77
3.3 Print out from spectrum analyzer. 86
3.4 Experimental arrangem ent for acquisition of the coil map. 88
4.1 Summed patient spectra of 5FU. 106
4.2 2D stacked patient spectra of 5FU. 107
4.3 Plot of 5FU spectra. 107
4.4 Bargraph of data in table 4.1. 108
5.1 Structure of Gd-DTPA. 119
5.2 Graph of patient DEMRI (120 pts). 122
5.3 Graph of patient ICAR. 124
5.4 Graph of patient DCAR. 125
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
v ii
5.5 W olfs 3 compartmental model. 126
5.6 Axial MRI view of patient through the liver with tumors. 133
5.7 3D view of selected area of the tumor. 134
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Vlll
LIST OF TABLES
Table List of Tables Page
1.1 Pharmacokinetic parameters of 5FU administered as an
iv bolus injection. 24
1.2 Resistance to 5FU. 31
1.3 Factors leading to plasma drug concentration being a poor
predictor of therapeutic success. 39
1.4 Limitations of “classical” pharmacokinetic modeling for
cancer therapy. 41
1.5 Properties of solid tumors that differ from normal tissue
and can affect transport. 44
1.6 Various mathematical modeling approaches to plasma
concentration of drug. 54
4.1 Association of tumor trapping of 5FU with patient response
to chem otherapy. 108
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Abstract
To improve diagnostic and chemotherapeutic evaluation of solid tumors in
patients non-invasively a series of surface translocatable (STL) radio
frequency coils were designed, tested and manufactured in-house. The STL
coils were used in the 1 9 F spectroscopy (5-fluorouracil (5FU)) experiments
at several medical imaging centers on over 100 patients. This improved
protocol permitted a dramatic reduction in patient contact time, and thus
enhanced patient comfort, as well as allowing smaller doses of the
chemotherapeutic agent to be used. 71 % of the patients were deemed to be
5FU trappers (p = 0.0001), but 41 % of these were non-responders. The
oncologists were able to use this data to determine whether patients
responded to 5FU administration in a shorter period of time. Dynamic
enhanced magnetic resonance imaging (DEMRI) with a surrogate marker,
gadolinium diethylene triamine pentaacetic acid (Gd-DTPA) was effected
on 30 patients to determine the degree of vascularization of the tumors.
This was achieved by the method of initial contrast accumulation rate
(ICAR) measurements. In addition these experiments led to a novel method
for determining relative ion osmotic pressure of the tumor interstitium
which involved the process of delayed contrast accumulation rate (DCAR)
measurements. However, the latter results must be regarded only as
preliminary. The indications are that DEMRI may be a powerful means of
determining the extent of neo-vascularization and characteristics of
tumors. With further work this may lead to a better tool for testing tumor
trapping in vivo without the use of toxic chemotherapeutic agents.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1
SECTION 1 (GENERAL INTRODUCTION)
1.1 TUMOR GROWTH AND ANGIOGENESIS
Cancer is the second leading cause of death in the United States.
Cancer is not just one disease; rather, it is a general term for a number of
conditions characterized by uncontrolled cell growth. One group of
tumors grow as solid masses (e.g. tumors of the lung, bladder, breast,
colon) and studies measuring aspects of the perfusional properties of such
tumors will be the main thrust of work presented here.
Solid tumors, which start with a single or a few cells, grow into
significant solid masses imbedded in other tissues. While the first few
cancer cell use the existing vascular supply, as the tumor mass grows
these tumors require neovascularization.
1.1.1 PREVASCULAR PHASE
In the prevascular phase tumors generally grow in thin sheets of
one-to-two layer cells, and the number of cells is small, even less than 100.
Angiogenesis is absent, and the tumor population precludes expansion in
spite of its proliferative capacity. The tumor cells can escape
requirements of neovascularization, growing in these sheets as observed
in the meninges or inside nerve sheathes. The volume of tumor may
reach about 1 mm in this phase because all the necessary nutrients and
waste products are able to diffuse across a distance of 150 pm. Once this
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2
threshold size of growth is exceeded vascularization must commence to
allow exponential tum or growth and proliferation.
1.1.2 ONSETOFANGIOGENIC ACTIVITY
The onset of angiogenic behavior of tumor cells is an event
independent of expression or growth at quite different times during
neoplasia.
However, these examples do not necessarily imply that angiogenesis
originates only from tumor cells because other mechanisms also induce its
onset, as, for instance, by the recruitment of macrophages. Thus, tumor
cells must have the ability to induce neovascularization and growth by
means of appropriate internal factors and environmental conditions
specific to the location or organ. Paget (1889) suggested the latter over a
century ago in his “seed and soil” hypothesis.
Tumor angiogenesis begins with the growth of new capillaries from
preexisting microvasculature. The basement membrane of the host vessel
is degraded by enzymes provided by the tumor cells and the capillary
endothelial cells travel towards the tumor to form new capillaries under
the influence of angiogenic factors (Folkman and Klagsburn, 1987; Moses,
1991; Denekamp, 1993). A considerable number of such factors are known,
and at least 12 to 14 of these factors have been identified (Fox et al., 1993)
examined 20 invasive human breast carcinomas and found that endothelial
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3
cell proliferation occurred at the tumor periphery. When endothelial cell
division has been accomplished, restructuring and migration of the
existing tissue vascular supply occurs to provide the internal vasculature
of the tumor.
1.1.3 ANGIOGENIC DEPENDENCEOFTUMORGROWTH
That tumor growth is dependent on angiogenesis may be determined
from various kinds of evidence. For instance, tumors grown in isolated
perfused organs in which blood vessels do proliferate rarely exceed about
1 mm3 in volume but those tumors implanted into mice expand rapidly to 1
to 2 cm3 after vascularization (Folkman et al, 1987). Similarly, tumor
growth in the vascular cornea advances linearly and slowly until
vascularization takes place; exponential growth follows. Explosive growth
is a feature which generally follows angiogenesis.
1.1.4 TEMPORAL ONSET OF ANGIOGENESIS
Tumor growth may be expected to correlate with angiogenesis, and
should measurable by tumor blood flow. Two types of blood vessels make
up the tumor vascular supply: newly formed tumor vessels stimulated by
one or more tumor angiogenic factors, and normal vessels cooped from the
host tissue. The induced vasculature is generally abnormal and poorly
differentiated, lacking in smooth muscle cells. Such vasculature often
exhibits collapsed vessels because of increased tissue pressure, large
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4
sinusoidal structures, exteriovenous shunts, stasis and thrombosis (Vaupel
eta/., 1989).
Tumors become vascularized in mice after subcutaneous injection of
tumor cells when they reach about 0.4 mm3. As the tumor size increases,
the vasculature represents about 1.5% of the tumor volume 400% more
than that of normal subcutaneous tissue. Direct experimental evidence
exists for the correlation of the time of onset of tumor growth and
angiogenesis. For example, human colon carcinomas lack high-affinity
receptors for basic fibroblast growth factor (BFGF), and for which BFGF is
not mitogenic in vitro, but when the tumor was grown in nude mice,
systemic (intraperitoneal) injection of BFGF stimulated blood vessel
density and branching in the tumor which doubled in size. On the other
hand, injection of neutralizing monoclonal antiserum to BFGF
significantly slowed tumor growth, confirmed by receptor
autoradiography of histological tumor sections which showed that these
receptors were located on the vascular endothelium.
Angiogenesis also contributes to metastasis. Neovascularization
allows cells to be shed from the primary tumor. Decreased angiogenesis
correlates with a decreased rate of metastasis.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5
1.1.5 TUMORS: STRUCTURE AND ITS ENVIRONMENT.
The tumoral blood space is made up of leaky vessels. Hence, the
fraction of blood that will flow through the blood vessels of a given tum or
mass will be a function of the branching ratio:
[volume of blood to the tumor]
Ratio=_______________________________________ Equation 1.1
[volume of blood flowing to rest of body ]
Also, the volume of blood flowing through a tumor is determined by the
number of blood vessels, and their density is a function of
angiogenic/antiangiogenic factors operating at that tumor site (Jain, 1988;
Vaupel et al., 1989). The interstitial fluid space is a key and a rather
unique element in solid tumors. Solid tumors lack a lymphatic drainage
system, and hence, lack the regulation afforded to them by homeostasis.
Pressures in solid tumors can vary dramatically, and will affect the
diffusion and the convection of products across that barrier.
Bhujwalla et al., (1992) suggest that solid tumors may be divided into
3 compartments consisting of tumor cells, the vascular compartment and
an interstitial compartment. Except for high lactate concentrations (10-30
mM) and low free glucose content (0-2 mM) (Less et al., 1992; Wiig, 1982;
Jain, 1987), the composition of tumor interstitial fluid is closely similar to
normal subcutaneous interstitial fluid. Furthermore, clinical and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6
experimental tumors usually exhibit higher interstitial fluid pressure
than that found in normal tissues. Such hypertension is consonant with
the absence of functioning lymphatics, a high filtration coefficient and
permeability of tumor vasculature, as well as the rapid proliferation of
tumor cells in confined spaces. Tumor endothelial cells double their
numbers in about 2-13 days compared to normal tissue endothelial cells,
which have doubling time of 47-200 days (Tannock, 1970; Hirst eta/., 1982).
Thus, tumor endothelial cells are often unable to match the rapid
proliferation of tumor cells. These leads to a reduced and inhomogeneous
blood supply, as well as insufficient substrates and oxygen, which, in turn
leads to hypoxia, anoxia and cell death. Thomlinson and Gray (1955)
showed that necrosis in human bronchial tumors took place at an
approximate distance of 150 |j.m from the nearest blood vessel which
suggested that the oxygen or substrate supply, or both, did not extend
beyond that range. These authors suggested the presence of viable
hypoxic cells in areas between the necrotic and the well-oxygenated cells.
Such hypoxic cells may survive radiation therapy and allow the tumor to
recur, hypoxia being “chronic” or diffusion limited. However, necrosis is
not invariably due just to the absence of capillaries or the absence of a
penetrating vascular bed in solid tumor cores as demonstrated by
metastatic ovarian carcinomas which have a pronounced border of vessels
in and around necrotic areas. In the latter case vascular collapse or acute
hypoxia may cause this effect. Falk advanced the notion that this anomaly
was due to heterolysis as opposed to autolysis. Heterolysis is a term
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7
introduced by Thomlinson to describe destruction and death caused by
agents produced at a distance. Weinhouse (1976) postulated that fatty acids,
proteins and nucleic acid fragments, and cellular debris found in necrotic
areas can also interfere with mitochondrial functioning of cells in areas
contiguous to necrotic regions.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8
1.2 BLOOD FLOW AND OXYGENATION
Within tumors blood flow and oxygenation in the physiological
environment are critical factors either directly or indirectly. Vascular
collapse and tumor necrosis has been observed frequently for impaired
blood flow and substrate. Because blood flow, oxygenation and metabolism
are closely linked, noninvasive NMR spectroscopic methods which
measure metabolism have been widely applied to the study of solid tumors.
MR imaging, also frequently applied, has the great advantage of
noninvasively determining tumor volume localization, the morphology of
solid tumors and tumor perfusion. Several good reviews of MR studies and
cancer are available. To increase the effectiveness of treatm ent and
diagnostic strategies monitoring blood flow and oxygenation allows
greater flexibility with non-invasive methods. Gray and coworkers (1953)
showed that oxygenation was crucial in the survivability of cells and
tissues to radiation damage. Thomlinson and Gray (1955) suggested that
cells suffering from hypoxia eventually become necrotic. Specifically
killing hypoxic cells in tumors has greater therapeutic potential than
oxygenating or chemically sensitizing them to radiation or chemotherapy,
since hypoxia is tumor specific (Graeber etal., 1996). Hypoxia has been
noted to be a common feature of human and animal tumors, and the vast
majority of human solid tumors have median pC>2 levels lower than the
normal tissue of origin, while in animal tumors hypoxic cells are viable
and contribute to the resistance of transplanted tumors to both radiation
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9
and some anticancer drugs. The classic model of Thomlinson and Gray
(1955) involves the diffusion distance of oxygen to hypoxic cells, but
another model advanced by Baxter el al, (1995) suggests that tumor
hypoxia can also occur by the temporary obstruction cessation of tumor
blood-flow leading to acute hypoxia. The latter model is believed to be
present in human as well as in animal tumors. Knowledge of blood flow to
and through tumors is thus of fundamental importance and in this respect
magnetic resonance techniques for researching these phenomena are of
the greatest utility, and also to predict the kinds of chemotherapeutic and
diagnostic protocols to apply to effect cures. Having illustrated the
importance of one of the key physiological aspects of tumors (blood supply
and vasculature), what would be the effect that chemotherapy could have
on tum or vasculature? 5-fluorouracil (5FU) is a widely used agent in the
treatment of solid tumors, allowing us to study the effect of this drug on
vascular changes in tumors.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10
1.3 PAST AND PRESENT USE OF 5FU
1.3.1 5-FLUOROURACIL (5FU)
5FU is an antimetabolite widely used in the treatm ent of solid
tumors, especially colorectal cancer. Its proposed mode of action is to act
competitively with uracil, one of the normal bases. 5FU will incorporate
into RNA in lieu of uracil, will bind into a ternary complex (and thereby
prevent the formation of thymine in tumor cells), and will incorporate
into DNA in lieu of thymidine.
The presence of the fluorine atom on this anticancer agent makes it
possible to measure this drug noninvasively using 1 9 F-NMR. In addition, a
good number of fluorinated compounds have been used extensively and
effectively as therapeutic and diagnostic agents, particularly fluorine
analogs of natural products. Since its discovery by Heidelberger in 1975,
5FU has been used in the treatm ent of solid tumors of the gastrointestinal
tract, head, neck and breast. 5FU can be administered intravenously as a
bolus or by continuous perfusion.
1.3.2 FLUORINE AND 5FU: STRUCTURE AND PROPERTIES.
1.3.2.1 FLUORINE
1 9
The stable isotope of fluorine F found naturally has an atomic mass
of 19 a.m.u., and an atomic number of 9. The fluorine van der Waals radius
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
11
is 1.35 A similar to that of the hydrogen atom, 1.2 A. For this reason
replacement of a hydrogen atom in a drug molecule by fluorine, as for
instance the hydrogen atom in the 5-position of uracil to yield 5FU, causes
minimal. Like hydrogen, fluorine forms only single bonds in organic
molecules, and all fluorinated compounds can be represented by the
general formula RF, where R may be an aliphatic, aromatic or
heterocyclic residue.
The fluorine atom is unique because it may be detected by a num ber
1 9
of sensitive techniques: the natural F isotope by NMR spectroscopy, and
1 o
the radioactive isotope, F (t1 /2 = 108 min, positron-emitter) by nuclear
1 9
imaging procedures, F has a spin number of 1/2, a natural isotopic
abundance of 100%, a gyromagnetic ratio of 1 His 2.79268 X104 rad per sec.,
e.g. is 42.6 MHz/T, while F is 40.05 MHz/T), 84% of a proton sensitivity.
1 9
F also has a large chemical shift range, approximately 200-ppm. For this
reason, in NMR spectroscopy, it may be used as a non-invasive diagnostic
tool in preclinical and clinical experiments.
1.3.2.2 5FU
5FU with a molecular weight of 130.08 is the analog of the naturally
accruing pyrimidine, uracil (Figure 1.1).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Structure o f 5FU and uracil
H\
H
H
I
H
C C l. lA f A l IPO/tl!
J - l I U W I W U I C I V I I
Figure 1.1
i
u iao u
Substituting the 5-H in uracil with F does not impede the conversion of 5FU
to nucleosides and nucleotides. However, the alicyclic F-C bond is very
stable and exhibits little chemical reactivity (Heidelberger 1975). 5FU is a
white to almost white crystalline powder sparingly soluble in water. As an
antineoplastic antimetabolite it is administered intravenously in the form
of a sterile, nonpyrogenic solution. However, the pH of the solution
requires adjustment to 8.4 to 9.2 with sodium hydroxide. The almost 0.4 M
concentration of the therapeutic 5FU solution is (0.3843 M, 500 mg/10 mL).
1.3.2.3 BIOCHEMISTRY OF 5FU
Uracil in a normal biochemical cycle is anabolized to deoxyuridinic
by the enzyme thymidine phosphorylase. Thymidine kinase thereafter
converts the substrate to deoxyuridine monophosphate, which is then
metabolized to deoxythymidine monophosphate by thymidylate syntheses.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13
5FU can also be incorporated into RNA instead of uracil, and into DNA
instead of thymidine. These processes all require, that 5FU administered in
vivo reach the target site. Such targeting involves transport and
metabolic changes. In the anabolic pathway, the metabolism of 5FU blocks
the methylation reaction of deoxyuridylic acid to thymidilic acid. Thus,
5FU interferes with the synthesis of deoxyribonucleic acid (DNA), and to a
lesser extent inhibits the formation of ribonucleic acid (RNA). DNA and
RNA are essential for ceil growth and division, so that the effect of 5FU
may result in a thymine deficiency to provoke unbalanced growth and
death of the cell. In those cells which form more rapidly, and which
therefore incorporate 5FU at a faster rate, the effects of DNA and RNA
deprivation are the most marked. To exert its effect 5FU requires
conversion to the nucleotide level (see figure 1.2). It can be incorporated
into RNA interfering with the maturation of nuclear RNA. However, its
main mechanism of action on DNA synthesis is by the conversion of 5F to
5-fluoro-2’-deoxy, 5-monophosphate (FdUMP) leading to inhibition of
thymidylate syntheses (TS). If a folate cofactor is present a ternary
complex is formed involving 5FU, the stability of which is the main
determinant of the action of 5FU. Diasio (1989) pointed out that the
anabolism of 5FU could occur through 3 different pathways. Anabolism of
fluorouracil to the cytotoxic nucleotides can occur through three different
pathways (Diasio 1989):
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
14
Dihydrofluorouracil
(DHFU)
► A -a f
A a ;
4
5-Fluorouracil (5-FU)
o
CO O H
6 h f
-
|(|H
6 0
N H ,
CO O H
6 h f
9H 2
n h 2
Fluoro-beta-alanine
(FBAL)
Catabolites
Fluoro-ureido-propionic acid
(FUPA)
5-Fluorouridjne f
(FUR)
Anabolites
\
\
5-Fluorodeoxyuridine m
(FdU R) ”
FUMP ■
FdUMP
FU D P-
FdU D P
Thymidylate
Synthase
Catabolites and Anabolites of 5FU
Figure 1.2
(1) 5-Fluorouracil reacts with deoxyribose-1-phosphate in the presence of
thymidylate phosphorylase producing fluorodeoxyuridine (FdURD) which
is then converted to fluorodeoxyuridine monophosphate (FdUMP) by the
enzyme thymidine kinase. (Woodman 1980)
(2) 5-Fluorouracil and ribose-l-phosphate, in the presence of uridine
phosphorylase, is converted to fluorouridine (FURD). Phosphorylation of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
15
FURD by the enzyme uridine kinase yields fluorouridine monophosphate
(FUMP).
(3) 5-Fluorouracil reacts with phosphoribosyl pyrophosphate (PRPP) to
give FUMP. This is a one step reaction catalyzed by phosphoribosyl
transferase (PRTase), (Kessel 1972) specifically, orotic acid phosphoribosyl
transferase (OPRTase).
Sequential action of uridine monophosphate kinase and uridine
diphosphate kinase on FUMP results in the formation of fluorouridine
diphosphate (FUDP) and triphosphate (FUTP) respectively. This fraudulent
nucleotide, FUTP, is incorporated in the RNA by the enzyme RNA
polymerase. The fluorouridine diphosphate (FUDP) can also be converted
to the deoxyribonucleotide, fluorodeoxyuridine diphosphate (FdUDP) by
ribonucleotide reductase (Kent 1972). FdUDP can either be phosphorylated
to FdUTP or dephosphorylated to FdUMP. Uridine monophosphate kinase
and diphosphate kinase to yield the diphosphate (FdUDP) and the
triphosphate (FdUTP), respectively phosphorylate fluorodeoxy uridine
monophosphate (FdUMP). DNA polymerase, which cannot distinguish
between thymidine triphosphate (TTP) and deoxyuridine triphosphate
(dUTP), also accepts FdUTP and incorporates it into the DNA chain,
inhibiting further elongation of the DNA chain. Contrary to earlier belief
that fluorouracil only acts on the RNA, recent studies report the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
16
incorporation of intratumoral 5FU into the elongating DNA chain. (Kufe
1981).
All the three activation pathways of fluorouracil described earlier
offer cytotoxic efficacy. But it is not certain which of these three is the
primary or most effective mechanism of 5FU activation. The direct
pathway via orotic acid phosphoribosyl transferase has been predicted to
be the most predominant route of activation in several different murine
leukemia such as P1534, P338, and Sarcoma 180. (Kessel 1972, Reyes 1969).
The indirect pathways via uridine phosphorylase and uridine kinase have
been reported to be more common in the Novikoff hepatoma, Walker 256,
HeLa and human colon carcinomas grown in nude mice. (Houghton 1980,
1983, Benz 1981). The initial metabolism of 5FU to nucleotides such as
fluorouridine-5-triphosphate (FUTP) and 5-fluoro-2-deoxyuridine-5-
monophosphate (FdUMP) is essential for its action. Several enzymes
belonging to pyrimidine metabolism are required for the conversion of
5FU to nucleotides. FdUMP can be formed from FUMP via reduction of FUDP.
The extent of growth inhibition by 5FU may be correlated with the activity
of one or more of the enzymes catalyzing the initial metabolism of 5FU. For
some cell lines orotate phosphoribosyl-transferase (OPRT) has been shown
to play a major role in the initial metabolism, whereas for other cells
uridine phosphorylase is more important. 14 Nucleotides formed via the
direct pathway (via OPRT) and the indirect pathway (via FUR) are
incorporated into different RNA fractions. Sensitivity of cell lines and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
17
tumors to 5FU might also depend on the availability of co-substrates
required for the conversion of 5FU to active nucleotides. The catabolic
metabolism of 5FU yield degradation products, amongst other: carbon
dioxide, urea and 2-fluoro-ft-alanine, these are inactive.
5-Fluorouracil is catabolised to dihydrofluorouracil (DHFU) in the
presence of dihydrouracil dehydrogenase (DHUD) (Naguib 1985). This
unstable intermediate is spontaneously hydrolyzed to fluoroureido
propionic acid (FUPA) by dihydropyrimidase (Maguire 1978). Finally
ureidopropionase degrades FUPA to fluoro-p-alanine (FBAL), ammonia,
urea and CO2. (Figure 1.2) (Sommadossi 1982).
1.3.3 MECHANISM OF ACTION OF 5FU AS AN ANTITUMOR AGENT.
1.3.3.1 INHIBITION OF THYMIDYLATE SYNTHETASE (TS)
The best studied, understood and long recognized mechanism of 5FU
action is by the inhibition of TS by FdUMP. (Danenberg 1977, 1982, Santi
1974). In a mammalian cell, dUMP is converted to dTMP by thymidylate
synthase. During the methylation of dUMP, the folate group is attached to
the 5-position of the pyrimidine ring. The methyl group is donated by
5,10-methylene-tetrahydrofolate (MTHF). (Figure 1.3). The same reaction
occurs with FdUMP but the ternary complex formed with FdUX does not
proceed any further (Danenberg 1978, Santi 1972).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
18
5, 10-Methyltetrahydrofollo acid
H
CH
OH
HN
■C-NH-CH
CH
COOH
FdUMP
HO-R-O-CH.
OH
OH
Ternary complex of 5,1O-methylene-tetrahydrofolate
Figure 1.3
For dUMP, the methylene group is reduced to a methyl group by removal
of a proton from the 5-carbon position, forming an intermediate, which
can act as a hydride receptor in the reduction reaction. (Danenberg 1982).
FdUMP shows a higher binding affinity for TS than natural substrate of
DUMP. The fluorine on the 5-position of the pyrimidine ring of 5FU
prevents this intermediate to form, terminating the action of thymidylate
synthase at this point. Thus the enzyme is trapped in a ternary complex
form with FdUMP creating a 'thymine-less' state inside the cell which is
toxic for actively dividing cells. The covalent bonds between FdUMP and TS
and between FdUMP and 5,10 methylene tetrahydrofolate are chemically
stable and irreversible.(Santi 1974). Dissociation of FdUMP and MTHF
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
19
results in the free active enzyme again that can take part in normal
reaction. In the presence of a reduced folate cofactor, the binding is
tighter and this inhibits the dissociation of the ternary complex. (Ulman
1978). Based on this report, a modulation with a reduced folate cofactor
example, folinic acid (Halaka 1984), enhances the antitumor effectiveness
of 5FU. The higher the concentration of reduced folate, the tighter are the
bonds in the ternary complex and the more difficult the dissociation.
(Evans 1984). Methotrexate acts in a similar manner by increasing the
strength of binding, but compared to a natural folate, Methotrexate binds
more loosely in complexes and can more readily dissociate.
1.3.3.2 INCORPORATION OF FdUTP INTO DNA
Contribution of 5-fluorouracil cytotoxicity by incorporation into
DNA had been considered to be an unlikely event for a very long time.
Recently, a possible mechanism for such incorporation has been
identified. Following anabolism of 5FU, DNA polymerase (Ingraham 1980)
acts on the anabolite, FdUTP, which is then incorporated in the DNA chain
in place of dTTP. (Tanaka 1984, Scheutz 1984, Sawyer 1984). Two enzymes
prevent incorporation of the fraudulent nucleotides FdUTP and dUTP into
the DNA. First, dUTP hydrolase, a phosphorylase, cleaves dUTP and FdUTP to
dUMP and FdUMP respectively. (Figure 1.4). Second, Uracil DNA
glycosylase catalyses the hydrolysis of fluorouracil-deoxyriboseglycosyl
bond of FdUMP residues present in the DNA; (Ingraham 1980) its activity
correlates inversely with the level of FdURD incorporation into the DNA.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
20
(Tanaka 1984). The endonucleolytic cleavage of the base-free deoxyribose
site results in a strand break, which is repaired subsequently.
dGTP — — ►
D
dATP
-----► N
dCTP
-----► A
dUMP
dUDP UTTP
dTMP _► dTDP
dTTP
FdUMP
dUDP dUTP
Effect of 5-fluorouracil on DNA synthesis.
Figure 1.4
Incorporation of FdUTP into the deoxyribonucleotide chain alters DNA
stability producing small fragments of DNA strands. (Cheng 1983).
Treatment with 5FU causes DNA chains termination and also reduces the
average chain length of the DNA (Scheutz 1985). There is no well-
established correlation between the incorporation of 5FU into DNA and
cellular toxicity. Some DNA repair does occur following incorporation of
5FU into DNA but in the presence of considerable damage, such repair may
not be effective and also leads to fragmentation. (Scheutz 1955). Some
observations suggest that the deoxyribonucleotide triphosphate imbalance
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
21
produced by FdURD triggers synthesis of an endonuclease capable of
inducing strand breaks. (Yoshika 1987).
1.3.3.3 INCORPORATION OF FUTP INTO RNA
5-Fluorouracil is extensively incorporated into cytoplasmic and
nuclear RNA fractions altering RNA processing and functions as well as
cell viability. It is incorporated into each species of RNA including
ribosomal, messenger and transfer RNA after it gets inside nuclear RNA
(Cohen 1985, Armstrong 1986). First, fluorouracil metabolized to FdUMP or
FUTP is incorporated into the nuclear RNA. The cytotoxicity of 5FU is
usually determined from the extent of its incorporation into nuclear RNA
although the specific molecular locus for cytotoxicity remains to be
clarified. Once it is placed inside the nuclear RNA, 5FU inhibits the
processing of nuclear RNA to low molecular weight ribosomal RNA,
especially the transition from the 45s to the 28s and 18s units. (Kanamaru
1986) Fluorouracil exposure of cells also leads to the inhibition of transfer
RNAuracil-methyltransferase resulting in modified uridine bases in
transfer RNA (tRNA). It can cause irreversible inhibition of RNA
methylation. (Santi 1987). Fluorouracil also blocks the processing of
nuclear messenger RNA (mRNA) coding for dihydrofolate reductase
(DHFR) and decreases polyadenylation of this species. (Armstrong 1986).
Summarizing the above discussion; it is evident that misincorporation of
bases into nuclear RNA is associated with a block in processing and/or
nuclear cytoplasmic transport of other RNA's, e.g., tRNA, mRNA.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
22
1.3.3.4 ALTERATION OF CELLULAR MEMBRANES AND DECREASED
GLYCOPROTEIN SYNTHESIS
Although the majority of the 5FU induced cytotoxicity can be
attributed to its inhibition of the thymidylate pathway and altered RNA
function, part of 5FU's action is also mediated through its interference
with other cellular processes. 5-fluorouracil changes the transmembrane
potential and decreases surface charge on the membrane of tumor cells,
(Walliser 1978) interrupts glycosylation of proteins and lipids necessary
for cellular functions involving cell surface glycoprotein and glycolipid
receptors causing abnormal structural changes in the cell membranes.
Such membrane impairment is associated with an increase in tum or cell
volume, making the cells highly susceptible to lysis and destruction.
(Kessel 1980)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
23
1.4 PHARMACOKINETICS (PK) OF 5FU
The pharmacokinetic of 5FU has been reviewed extensively by
others (El Sayed and Sadee, 1983). Initial assays of 5FU lacked either
specificity or sensitivity (El Sayed and Sadee, 1983). Currently, the most
widely used method is high performance liquid chromatography (HPLC)
combined with UV absorption with a detection limit of 0.5 to 1.0 gmol/L 5FU
(Leyva et si., 1984; El Sayed and Sadee, 1983; Peters et si., 1984) A lower
detection limit for 5FU (down to 3 x 10-9 mol/L; 0.3 ng/mL) can be achieved
with gas-chromatography-mass-spectrometry (GC-MS)" (Finn and Sadee,
1975; Kok et al., 1985). Most of the pharmacokinetic studies were restricted
to three hours or less, but with the sensitive GC-MS method 5FU plasma
concentrations can be followed for at least eight hours after the
injection," which makes it possible to perform long-term pharmacokinetic
studies (Van Greningen et al., 1989). The pharmacokinetic of single-dose
5FU administered as an intravenous (iv) bolus injection in doses ranging
between 300 and 600 mg/mm2 have been studied in detail (Meyers, 1981;
Collins et al., 1980), and the findings are summarized in Table 1.1. Rapid
distribution over a large volume and rapid elimination have been reported
to follow peak levels lying in the millimolar range (Figure 1.5). The total
clearance was rather high (Table 1.1),
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
24
Table 1.1 Pharmacokinetic Parameters of 5FU
Administered as an iv Bolus Injection.
Parameters Value
Peak levels 10-4 - 10-3 mol/L
Volume of distribution 8 - 54 L
Total clearance 0.5 - 2.0 L/min
N ,
r*J
0 l
2 }
hr
Curves for 5FU and F-DHU Post-iv Bolus Injection
Figure 1.5
and comparable to the liver flow, but hepatic extraction has been
estimated to be 50% (Ensminger et al., 1978), yet the liver is the organ with
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
25
the highest level of dihydrouracil dehydrogenase activity (Youngner et
al., 1979). The kidneys, in which the activity of this enzyme is also high,
contribute to elimination of 5FU by both degradation and active renal
excretion, about 20% being excreted as the parent drug (El Sayed and
Sadee, 1983). The lungs have also been reported to be a major site of 5FU
clearance (Meyers, 1981; El Sayed and Sadee, 1983; Collins, 1985),. Collins et
a l, have shown that a saturable two-compartment model can be used to
describe the elimination kinetics of 5FU. Calculation gave an apparent Km
of 15 pmol/L in plasma.
Nonlinearlity of 5FU kinetics has been described by several authors
(De Meyers, 1981; El Sayed and Sadee, 1983), and is related to the saturation
of 5FU catabolism. Studies on the pharmacokinetics of 5FU catabolites have
been hampered by the lack of appropriate detection methods. A relatively
insensitive method applied HPLC (Figure 1.5), and a more sensitive method
19
used GC with electron capture detection. With F-NMR, the other
catabolites could also be demonstrated in human plasmas Analysis of the
cumulative urinary excretion of these catabolites showed that F-BAL was
the major one followed by FUPA. F-DHU was a minor constituent of the
urinary excretion products (Malet-Martino et al., 1985). The high
detection limit of 10 gmol/L, which is also in the range of the peak plasma
concentration (Malet-Martino et al., 1985), is the major limitation of this
technique. Improvement might permit investigation of the dose-
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
26
dependency of 5FU pharmacokinetic in humans in relation to the in vivo
behavior of F-DHU.
There is no evidence that continuous iv administration of 5FU is
associated with a higher antitumor efficacy than bolus administration.
These two schedules give quite different types of toxicity, mucositis being
dose-limiting for infusion and myelosuppression for the bolus injection.
The pharmacokinetics of continuous 5FU infusion differ significantly
from those of the iv bolus, the former having a much higher clearance
value of 2 to 6 L/min (Collins et al., 1980), which considerably exceeds the
hepatic flow of 1.5 L/min to approach the cardiac output. This high
clearance level can be explained mainly by the high pulmonary
extraction (El Sayed and Sadee, 1983; Collins et al., 1980; Collins, 1985)
Pulmonary extraction accounted for a clearance higher than the cardiac
outputs. However, it has been shown that the liver and kidneys also
contribute to clearance.
In the past 5FU was administered orally. However, there is a marked
variability in its bioavailability, ranging between 28% and
100% (Laufman et al., 1987; Phillips et al., 1980). This finding may be
related to a saturable hepatic metabolism induced by dihydrouracil
dehydrogenase (Collins et al., 1980) (Figure 1.2), but also to an additional
first-pass effect arising from the rather high mucosal activity of
dihydrouracil dehydrogenase. Because of the substantial variability
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
27
observed, it is generally accepted that 5FU should not be administered
orally.
5FU is also administered intrahepatically by portal or arterial
infusion for the treatm ent of liver metastases. Specifically for this route of
administration, hepatic extraction and the rate of infusion determine the
systemic availability. The use of rapid intrahepatic arterial infusions at a
high dose (1,000 mg/m2 /d ) gave relatively low hepatic extraction
amounting from 20% to 60%, which led to a high systemic availability.
With a slower infusion rate and/or lower doses (780 mg/m2 /d), hepatic
extraction exceeded 90% (Ensminger et al., 1978; Boublil et al., 1985) and
this was accompanied by low systemic toxicity. More evidence pointing to
hepatic saturation was provided by the observation that 5FU levels rose
significantly during the infusion (Boublil etal., 1985). This new
pharmacokinetic information makes it possible to design better 5FU
schedules for the treatm ent of liver metastases.
Intraperitoneal infusions offer the possibility of achieving higher
drug concentrations, they give optimal exposure of tumor tissue within
the abdominal cavity, and provide more effective treatment of not only the
liver (via the portal vein) but also of peritoneal metastases. 5FU can be
administered by intraperitoneal peritoneal dialysis (Speyer et al., 1980) or
via implantable devices (Gyves, 1985). Intraperitoneal 5FU was cleared at a
rate of 14 mL/min, and 82% of the 5FU administered was absorbed within
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
28
four hours. Hepatic extraction was calculated to be 67% (Speyer et al.,
1980). A 2- to 3-log difference was observed between peritoneal and plasma
5FU concentrations (Speyer et al., 1980; Gyves, 1985). With continuous
infusion the mean steady-state level of 5FU in the intraperitoneal cavity
was 62.2 pmol/L (Gyves, 1985). Total body clearance ranged from 0.9 to 16.5
4 min Speyer et al., 1980), which is similar to the rate seen with continuous
iv infusion of 5FU. Clearance decreased with increasing 5FU
concentration, which is consistent with saturable or nonlinear 5FU
pharmacokinetic. It might be worthwhile to study this method of
administration in an adjuvant setting after surgical removal of Dukes B2
and C colorectal cancers.
1.4.1 INACTIVATION OF 5FU
5FU can be inactivated by degradation to a-fluorodihydrouracil (F-
DHU) (Figure 1.2). Further degradation of 5FU has been studied as far as to
a-fluoro- ft-alanine (F-BAL), and further metabolism of F-BAL is not
unlikely. fi-Alanine itself is a substrate for carnosine, but it can also be
converted to acetate. Similarly, F-BAL can be converted to fluoroacetate,
which has been related to neurotoxicity (Koenig and Patel, 1970). 5FU
degradation occurs in all tissues, but tumor tissue contains very small
amounts of dihydrouracil dehydrogenase (Ho et al., 1986). The activity of
this enzyme, while occurring in the kidney, is most intense in the liver
(Youngner et al., 1979), which means that the liver plays an important
role in 5FU degradation and elimination.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
29
It has been shown in patients that large amounts of 5FU are
19
degraded to F-DHU (De Bruijn et al., 1983), and fluorine-19 ( F) nuclear
magnetic resonance (NMR) has been used to show that, in vivo, FDHU is
rapidly degraded further to F-BAL(Malet-Martino eta?., 1987). Recently, a
new catabolite of 5FU was detected in bile, and was identified as an N-
cholyl-a-fluoro-^-alanine conjugate by NMR (Malet-Martino et al., 1987)
and enzymatic methods(Sweeny et al., 1987). In vivo inhibition of 5FU
degradation has been thought to increase the availability of 5FU to tumors.
However, it has been reported that an improved therapeutic index was not
observed clinically after the administration of thymidine and 5FU, but in
rats toxicity appeared to be increased (Woodcock et al., 1980). Impaired 5FU
degradation due to a deficiency of dihydrouracil dehydrogenase led to a
dramatic and fatal increase of 5FU toxicity (Engelbrecht et al., 1985).
It may be concluded that inhibition of 5FU degradation probably
will not improve therapeutic efficacy, since toxicity increases as much as
or even more than the antitumor activity.
1.4.2 DISTRIBUTION FOLLOWING iv INJECTION
7% to 20% of the parent drug is excreted unchanged in the urine in
6 hours; of this over 90% is excreted in the first hour. The remaining
percentage of the administered dose is metabolized, primarily in the liver.
The catabolic metabolism of fluorouracil results in degradation products
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
30
(e.g.,CO2, urea and a-fluoro-p-alanine) which are inactive. The inactive
metabolites are excreted in the urine over the next 3 to 4 hours. When
fluorouracil is labeled in the six-carbon position, thus preventing the 14C
metabolism to CO2, approximately 90% of the total radioactivity is excreted
in the urine. When fluorouracil is labeled in the two-carbon position
approximately 90% of the total radioactivity is excreted in expired CO2. 90%
of the dose is accounted for during the first 24 hours following
intravenous administration.
Following intravenous administration of fluorouracil. the mean
half-life of elimination from plasma is approximately 16 minutes, with a
range of 8 to 20 minutes and is dose dependent. No intact drug can be
detected in the plasma 3 hours after an intravenous injection.
1.4.3 RESISTANCE TO 5FU
5FU resistance, whether intrinsic or acquired, is usually caused by
aberrations (see Table 1.2) in the metabolism of 5FU or altered effects of
5FU metabolites. Generally, studies on 5FU resistance have been
performed by comparison of several tumor cell lines with different
sensitivity of 5FU or selection of a 5FU resistant subpopulation from a
sensitive tumor or cell line. It has already been shown by Reyes and Hall
(1969) and Kessel et al., (1966) that tumors with a low level of anabolism
have a low sensitivity to 5FU. OPRT may be the limiting enzyme for 5FU
anabolism (Reyes and Hall, 1969). 5FU transport across the cell membrane
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
31
might not limit its activity, but FUdR resistance was found to be related to
deficiency in its transport (Sobrero et al., 1985). Depletion of cosubstrates,
i.e., (deoxy)Ribose-l-P or PRPP, seems to limit the anabolism of 5FU, as
suggested by indirect evidence (Berger, 1983). Increased availability of
Ribose-l-P (Shani and Danenberg, 1984), deoxy-Rib-l-P (Evans et al.,
1981), or PRPP enhanced the sensitivity of 5FU. Enhanced nucleotide
catabolism due to a high level of alkaline phosphatase activity has been
shown to affect FUdR toxicity(Fernandes and Cranford, 1985). Altered
deoxythymidine triphosphate (dTTP) levels also affect FUdR toxicity
(Cohen and Ulman, 1978).
Table 1.2 Resistance to 5FU
• Deficiency of 5FU anabolism
• Deficiency of 5FU transport
• Depletion of essential cosubstrates
• Enhanced catabolism of 5FU, FUMP, or FdUMP
• Enhanced intracellular uridine concentrations
• Altered dTTP levels
• Alterations in thymidylate synthase
Aberrations in TS kinetics can lead to resistance against 5FU.
Several forms of aberration are summarized in Table 1.2. Altered enzyme
kinetics (Bapat et al., 1983) for TS were reflected by a higher dissociation
constant for the ternary covalent complex, but also by a weaker binding of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
32
dUMP. Intrinsic resistance to 5FU has been associated with high
accumulation of dUMP (Houghton et al., 1986). The turnover of TS was
higher in a resistant sub-cell line than in the sensitive cells (Bapat et a /.,
1983). A higher activity of TS was found in an FUdR resistant sub-cell line
(Priest el al., 1980), Possibly due to gene amplification. Amplification of
the gene coding for TS has been shown(Berger et al., 1985), recently also
in a case of human colon cancer with acquired resistance (Clark et al.,
1987). The stability of the ternary complex depends on the concentrations
of dUMP and FdUMP and the kinetic parameters, but also on the
availability of folates. Low total folate pools were associated with 5FU
resistance (Yin et al.,1983), as well as a low proportion of polyglutamate
derivatives (Yin etal., 1983). It may be concluded that resistance to 5FU
can be due to a variety of aberrations in 5FU metabolism, but factors
affecting TS appear to be of major clinical relevance.
1.5 CANCER M ODELS
For an anticancer drug to be effective on tumor cells it must reach
its target. In the case of radio-labeled drugs the isotope must be within the
range of the cells which for p-emitters is about a few hundred microns.
Any drug in this context is frequently injected or otherwise administered
in an area relatively distant from the targeted cells. In vitro or in small
test animals many anticancer drugs are quite effective but they fail in
human clinical trials. Moreover, such drugs show a wide variety of
responses in humans in spite of the same dosage, site of tumor and tumor
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
33
burden. Since most drugs act locally, the evidence for these differences in
response is the amount that reaches the target, rather than the tumor cell
response.
Interspecies or interpatient variation in drug delivery may be
explained in part, by the fact that human patients are larger than
conventional test animals but that the drug essentially still diffuses the
same distance in the same time (Strasser eta/., 1995). For instance,
xenograph tumors in mice are often of a single subcutaneous nodule
which may have completely different drug delivery patterns from widely
disseminated micrometastases of the same cell type in human patients. In '
addition many solid tumors are poorly vascularized making drug
penetration limited after intravenous injection (West etal., 1980; Sands e t
al, 1988). Hemingway et al., (1992) showed that colorectal liver metastases
are hypovascular and their perfusion is associated with limited drug
absorption and response, whereas head and neck cancers like squamous
cell carcinomas (Lamer etal., 1996) much better vascularized, show a
much better response.
The problems encountered with intravenous delivery of drugs to
solid tumors has received attention by Jain who have also recorded their
observations with respect to other routes of delivery. The importance of
drug delivery is still of frequently overlooked as can be gathered by a
recent review of pharmacokinetic optimization of chemotherapy by
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
34
Masson and Zanikou (Masson and Zamboni, 1997) who list factors
contributing to inter- and intrapatient variability of therapeutic success.
In addition to discussing the time history variability pharmacokinetic
(PK) they also refer to the reaction of an individual cell to the drug
pharmacodynamic (PD). Unfortunately these authors did not address the
relation between plasma and tumor concentration, i.e., delivery from blood
to tumor. In terms of delivery immuno liposomes have been developed
which theoretically bind to and kill individual tumor cells (Goren et al.,
1996; Ahmad and Allen, 1992), but experimentally the liposomes were
found usable to penetrate beyond tens of microns outside microvessels and
thus incapable of reaching the majority of cells in a tumor.
Increasing dosages to the maximum tolerable level by means of
improving host rescue techniques has led to better responses while
remaining within the lethal boundaries. The limits of improvement
achievable is being researched. However, the exact dosage given to a
patient has yielded better strategies in increased plasma concentration,
but not necessarily optimally so, as found in the case of doxorubicin. Every
increase in plasma concentration of a drug has produced clinically
significant increases in killing tumor cells for at least a proportion of the
patient, or cell, population. Yet for more recently introduced agents such
as monoclonal antibodies and liposomes, efficacious outcome has not been
improved by raising toxicity to the highest possible level (Vaughan et al.,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
35
1987). Thus therapeutists are now forced to investigate improved delivery
as an alternative route to regimen optimization.
1.5.1 K EY FACTORS
Improved quantitative understanding of processes involved in drug
delivery can be assisted by mathematical modeling which should take into
account two key factors: spatial variations in perfusion, drug
concentrations and micro structure in tumors, and non-equilibrium, time-
dependent aspects of drug delivery, cell kill and toxicity in normal tissue.
1.5.2 SPATIAL VARIATIONS
Gradients of drug concentration on several spatial length scales
variations across an entire tumor mass may be attendant on heterogeneous
perfusion and high interstitial pressure in tumor interiors. Drug
concentrations may differ widely over the distance between microvessels
depending whether interstitial transport is slow relative to transvascular
transport, or absorption and cell binding is relatively rapid. Orders of
magnitude differences occur for net rates of transport across cell
membranes, especially for resistant cells intracellular compared to
extracellular concentrations may be much lower, which is an example of
variation on cellular length scales.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
36
Tumor interiors often show substantially lower perfusion than in
growing tumor peripheries or in the contiguous normal tissue, this being
a major reason for the generally non-uniform drug concentration in
tumors (Endrich et al., 1979). The transport of iv injected drugs to adjacent
tissues is affected by perfusion, as is drug uptake by circulation for other
delivery pathways. Baxter and Jain refer to the fact that the absence of
lymphatics which are the cause of high interstitial pressure inside
tumors; this major factor leads to high interstitial spatial variations of
drug concentrations in a tumor. The model of Jain and Baxter (Jain, 1996,
1997; Goren et al., 1996; Endrich et al., 1979; Baxter and Jain 1989) clearly
exhibits the poor correlation between plasma concentration and drug
delivery to tumor cells, especially for relatively large drugs such as
monoclonal antibodies and liposomes (Baxter and Jain, 1989).
Experimentally determined large drugs reach only tum or peripheries or
regions contiguous to blood vessels, even vessels adjacent to tumors,
reported values of “percentage injected dose/gram of tumor” do not reflect
local concentrations experienced by all, even most tumor cells.
Many studies of anticancer drug distribution in the body concern
pharmacokinetic models (Ames et al., 1983; Strand et al., 1993), and as such
drug distribution to several body compartments is represented by a single
value of concentration gradients within each compartment, are not
considered explicitly, for the basic assumption is made that each
compartment is “well-mixed”.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
37
The simplest pharmacokinetic models consider only the drug
concentration in the plasma, frequently fitted to a bi- or tri-exponential
function. The exponential decays are rationalized to correspond with
distinct clearances mechanisms, but the connection between such
clearance rates and a given process of drug elimination in the body may
not be clear. More compartments are introduced in more complex models
(in physiologically-based models) which may correspond to organs or
tissues or to regions that equilibrate with plasma at different rates or
contain a metabolic product (in “operational models”) (Ames et al., 1983).
When a tumor is not always considered as a well-mixed compartment
communicating directly with the plasma, or as a separate compartment.
Such models, ignoring spatial variations in the tumor and other
body organs neglecting the spatially distributed nature of mass transfer
between plasma and tissue, may be named “classical” pharmacokinetic
models. These have contributed in the main to an understanding of the
relationship between injected dose and the temporal history of plasma
drug concentration. Relationships of this kind can vary considerably from
one drug to another, from patient to patient, from tumor to tumor types
because eliminating organs (Masson and Zamboni, 1997; Gamelin etal.,
1996; Balis et al., 1983) allows wide variations in drug clearance. Plasma
concentration is frequently a much more relevant index than injected
dose. Chemotherapy has been much improved by the monitoring of plasma
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
38
concentrations and consequently adjusting doses to sustain desired levels
(Masson and Zamboni, 1997) rather than injecting all patients with the
same dosage. Plasma concentration has been determined by extensive
experience to be the best predictor of toxicity to normal tissue (Masson and
Zamboni, 1997; Piscitelli etal., 1993).
Sader and Wong (1977) had difficulties correlating plasma
concentrations of 5FU with its therapeutic effectiveness. Thus, plasma
concentration has limitations as a measure of effective therapy. Muller e t
al., Muller et al., 1997 measured the area under the curve (AUC) of 5FU in
plasma and tumor interstitial fluid of breast cancer patients; tumor
response correlated with interstitial AUC but not with plasma. Table 1.3
lists a number of factors leading to plasma concentrations of drug being a
poor therapeutic predictor. Most of these factors are related to spatial
gradients in or around the tumor.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
39
Table 1.3 Factors Leading to Plasma Drug Concentration
Being a Poor Predictor of Therapeutic Success.
Factor
Explanation/ comments
Curative therapy needed
Killing only accessible cells is not curative
Unresectable bulky disease
No functioning vessels in interior
Large drugs cannot diffuse to interior
Large drug; diffusion across vessel wall is slow
Convective extravasation in tumor center
First agent m ay clear out o f tumor interior before second agent is
delivered
Small-molecular-weight drug
May achieve homogeneous distribution, but may also clear from
tumor faster
Drug acts only on cells in certain phase o f cell cycle (e.g., S or Gt)
Failure to kill dormant cells in interior
If plasma clearance fast relative to cell cycle, only fraction o f cells
killed
Drug requires good oxygenation for effectiveness
Failure to kill hypoxic cells in interior
One toxin molecule needed per cell killed
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
40
Drug m ust bind to all cells
Drug m ust be internalized by cell to kill
Drug m ust reach each cell
Binding site present only on fraction o f tumor cells
Toxic agent m ust kill tumor cells in vicinity also
Drug targets tumor vasculature
Nonfunctioning vessels in interior
Drug selectively extravasates in tumor
Failure to kill tumor cells migrating out into surrounding normal
tissue
In terms of compartments, a tumor and surrounding normal tissue
are regarded as distinct, separate units connected only by mass transfer
terms. But vanguard invading tumor cells which have migrated to
surrounding normal tissue, left untreated could eventually lead to relapse.
A theoretical study of Goitein and Schultheiss (1985) showed that outcome
is optimized when suspected surrounding tissue is treated with 70% dosage
given to the tumor proper. But drug development has been more focused
on exclusive delivery to the tumor itself, e.g., liposomes which escape
selectively from tumor cells but not from normal cells, and these drugs are
not likely to have access to isolated cancer cells. Current compartmental
models have difficulty in addressing this problem.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
41
Differences between a single bulky tumor and numerous
disseminated metastases are generally not accounted for by compartmental
models. Sgouros (1995) pointed out that for chemotherapy, as opposed to
radiotherapy response rates with bulky tumors could be used as predictors
of response for minimal residual disease.
Table 1.4 is a summary of some limitations of “classical” well-mixed
pharmacokinetic models for effectiveness of cancer therapies.
Time-dependent processes determine the distribution of a drug injected,
that throughout the body (and in the tumor, as well as killing of the tumor
cells and the toxicity to normal tissue. In this context the widest attention
has been said to be the time scale of plasma clearance. However, drug
extravasation, diffusion and convection through the interstitium,
transport across the cell membrane, uptake by microvessels, metabolism
by cells, clearance for the tum or and the other organs, and lethal action
on the cell are all processes generally not instantaneous compared to
plasma half life.
Table 1.4 Limitations of “Classical”
Pharmacokinetic Modeling for Cancer Therapy.
• Tumors (and other tissues) may not be well-mixed compartments
• Plasma may not be a well-mixed compartment
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
42
• Drug delivery to tumor may not vary in parallel with plasma
concentration
• Drug clearance from tumor may not occur in parallel with plasma
clearance
• Tumor cell kill may not correlate directly with local drug
concentration
• Mass transfer rates that are proportional to concentration difference
between compartments may fail to include significant convective
effects; conversely, those proportional to the concentration in one
compartment may neglect diffusive effects, or convection in the
opposite direction
• Modeling the tumor as a single well-mixed compartment fails to
distinguish between minimal residual disease, numerous disseminated
metastases, and large bulky tumor
• Isolated tumor cells invading surrounding normal tissue may not be
taken into account, but could be responsible for disease recurrence
• Area under the curve (AUC) as a measure of total exposure fails to
distinguish between low drug concentration for a long time and high
concentration for a short time, but the degree of cell kill may be quite
different
To kill a cell the temporal history of local drug concentration may
be more important than total exposure, measured by AUC, or area under
the curve. Bagudey and Finlay studied two anticancer drugs, taxol
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
43
(paclitaxel) and DACA for the same total exposure times, high
concentration for a short time is considerably less lethal than a lower
concentration for a longer time. On the other hand Nguyen-Ngoc et al,
(1984) showed that cultured mouse sarcoma cells exposed to doxorubicin
with low dose/long time incubation provide no cytotoxicity as high as that
obtained with high dose-short period approach. Juliano and Stamp (1978)
claimed that, “optimal therapeutic effects are obtained with S-phase
specific drugs when there is a continuous exposure of tumor cells to the
drug for a period of time greater than two generation times”. Failure to
distinguish these different penetration histories of exposure for values of
tumor blood or tumonnormal concentration ratios is evident in the
literature.
Instantaneous equilibration of tissue concentration with plasma
concentration is generally assumed for well-mixed compartmental
pharmacokinetic models. The well-mixed assumption is synonymous with
assuming instantaneous of the drug throughout the tissue interstitium.
Finite rates of transvascular and interstitial transport are considered
accounted for spatially distributed for interstitial concentration of the
drug, but they do not consider the rate of extracellular and intracellular
transport. Of course, cell kill at a given level of intracellular drug
concentration belongs to time-dependent processes.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
44
Theoretical models neglect many of the spatially varying and time
dependent factors not only in the neglect or averaging available data on
drug distribution. Swan (1988) states that, “The therapeutic objective is to
obtain the nature of the control agent [drug] that can drive the tumor
population to a desired level so as to penalize excessive usage of the drug”.
The connection between the obvious goal and commonly measured
pharmacokinetic quantities such as peak plasma concentration, area
under the curve (AUC) for plasma or tumor, or % injected dose/g in tumor,
liver, kidney, etc., at a chosen time of elapse after injection is unclear and
indirect.
An overview of existing theoretical models for drug delivery to
tumors will account for spatially heterogeneic and time-dependent factors
Our review is restricted to solid tumors - these being tumors currently
responsible for the majority of cancer deaths in the U.S., and such tumors
provide the most challenging obstacle to successful therapy. See Table 1.5
for special features.
Table 1.5 Properties of Solid Tumors That Differ from
Normal Tissue and Can Affect Transport.
• Absence of lymphatics
• Irregular vasculature
• Poor regulation of blood flow
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
45
• Heterogeneous perfusion; low in interior, higher than in periphery
• Lower pH than normal tissue
• High interstitial volume fraction
• High vessel hydraulic conductivity
• High vessel permeability; little dependence on molecular weight
• Different tissue elastic properties
• Higher tissue diffusivity and hydraulic conductivity
• High interstitial pressure in interior
• More hypoxic areas
• Different composition of extracellular matrix
Passive transport mechanism such as diffusion and convection
(solvent drag effect) are in the main responsible for drug extravasation or
absorption by vessels and for passage through the interstitium solid
tumors. Most tumors lack functioning lymphatics. For those that may have
some lymphatics active transport may take place for the drug from the
interstitium into the lymphatic system. We are concerned mostly with
passive transport systems.
Jain (1994) has indicated the clinical relevance of anti-cancer drug
distribution in terms of mathematical modeling. The relevance of this
approach has been heightened with the advent of more complex therapies.
Modeling can be a rapid and inexpensive way of assessing the effect of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
46
parameters. Where regression cannot be measured clinically, as in the
case of occult tumors, modeling is a rational approach to optimal therapy.
Basic knowledge of the complicated mechanisms of drug transport
within tissues, and their parameters, is lacking and this is the major
obstacle for mathematical or computational modeling generally require
simplifying assumptions and the estimation of key parameters. The
effectiveness of a model has a crucial dependence of the appropriateness
of such estimates and assumptions.
The administration of anticancer drugs by intravenous injection
(Balis et a/., 1983) is the most widely used method regardless of whether the
drug acts systemically or attacks a single large tumor mass (Jain,
1989,1994,1995,1996,1997) “Tissue-to-blood partition coefficient” (Chen
and Gross, 1979; Terasaki eta/., 1985) also known as “equilibrium
distribution ratio” is defined as the ratio of average tissue concentration to
outflow blood concentration (Farris et a/., 1988). As an expression of drug
uptake, this coefficient will be constant only if transport is “flow-limited”
(Farris eta/., 1988),i.e., if tissue concentrations are always equilibrated
with plasma concentrations. In actual cases, since the partition coefficient
varies with factors of tissue size and vascularization, its apparent value
also varies temporally for the same tumor. The latter is reported in many
experiments that the drug does not clear from the solid tumor hand in
hand with plasma clearance. Baguley and Finlay (1995) point out that in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
47
general tumors half-life clearance exceeds plasma half-life in this
context. They showed that in mice for instance, amsacrine, CI-921 and
DACA have a tumor half-life longer than that of the liver. For 5FU, Presant
and coworkers (1990) found a longer half-life in liver tumors for 6 out of
11 patients, and also for carcinomas of the pelvis, breast and lung
compared to plasma half-life by about 500%. Wolf et ai., (1990) Guerquin-
Kern etal., (1991) found similarly that the half-life elimination of 5FU
from tissue was about 35 min at pH 7.3, but about 90 minutes when the pH
was decreased to 6.9. The lower than normal pH of tumors may explain the
longer retention time. Blood stagnation was more prolonged in liver
tumors than in normal liver as Kameyama et al., (1989) showed. The tumor
must therefore be as a separate compartment, and that even for low
molecular weight drugs. Tumor concentrations cannot be assumed to
reflect directly plasma concentrations.
Since many anticancer drugs act only on the fraction of cells in a
particular phase of the cell cycle the long clearance time from tumors
compared with plasma may have a major effect on tumor cell kill. Such a
longer clearance time is an essential advantage when the drug is
administered with bifunctional antibodies or enzyme-conjugated
antibodies in two-step therapies (Baxter and Jain, 1996; Baxter etal., 1992;
Yuan etal., 1991).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
48
The assumption of uniform drug concentration throughout the
entire tumor compartment is a feature of many mathematical models
(Strand et al., 1993; Sung et al., 1994; Chen and Gross, 1979; Baxter et al.,
1995; Sung et al., 1992; Thomas et al., 1989). The difference between
average and local drug concentration in some regions of a tumor may
mean a significant difference in cell kill because many anticancer drugs
act locally.
The averaged Starling’s Law which treats that the tissue is
immediately adjacent to a vessel, may be used if the assumption that the
tumor is well-mixed on length scales typical of vessel spacing is valid.
Tissue-to-blood partition coefficients may be used if it is assumed that the
tumor tissue is equilibrated with the venous blood. In order that spatial
distribution of concentration can be replaced by the average
concentration one must assume that the entire tumor is well-mixed.
In a well-perfused tumor the diffusion times over a distance of 100
gm shows roughly how long the drug would take to reach cancer cells
furthest from the vessels. For the currently smallest clinically detectable
smallest tumor of 1 cm diffusion times for a distance of 0.5 cm indicate how
long the drug would take to reach cells in the center of such a tumor
should its interior not be perfused essentially. Swabb etal., (1974)
correlated tissue diffusivity as a function of molecular weight, but since
this correlation could not be based on adequate tumor data, it may be an
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
49
underestimate. Since tumors have a high interstitial volume fraction,
higher drug diffusivities in tissue might be expected compared with
normal tissue. Circulation time in plasma is not long enough for many
drugs to reach the center of very poorly perfused tumors. Effective
diffusion time will be increased by significant cellular uptake.
In the late 80s and early 90s while the issue of inadequate drug
delivery came to the fore with very disappointing results for large drugs
such as monoclonal antibodies and liposomes, there appeared evidence
that delivery of low-molecular weight drugs may also present a problem.
Even small chemotherapeutic drugs may vanish from circulation before
penetrating poorly perfused tumors. Antitumor response to
chemotherapeutic drugs also exhibit wide interpatient variability. Such
variability may in part be due to the extent of vascularization, although
interpatient variability research has focused mainly on multi-drug
resistance. Even well-vascularized tumors do not achieve uniform drug
distribution for high-molecular weight drugs, e.g., monoclonal antibodies
as shown by autoradiography which show the antibody only in regions
close to blood vessels.
Experimental observations that drug absorption is not proportional
to tumor mass suggests that tumors are not well-mixed compartments.
Williams et al, (1988, 1993) tried to fit data for tumor uptake (percent
injected dose/g tumor) for antibodies and liposomes as a function of tumor
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
50
mass and found a good correlation with a power law having an exponent
close to -0.33. This exponent was expected for a spherical tumor with the
drug being distributed uniformly in a spherical outer shell of thickness
independent of tumor diameter. These authors measured the absorption at
unspecified times greater than 24 h. In an earlier study, Redley et al.,
(1987) found a similar relationship for large drugs. An inverse
relationship between tumor uptake of antibody and tumor size was noted
by Cheung et al. (1997). Their absorption data plotted against tumor mass
on a logarithmic scale appear more or less linear, suggesting that g = A log
m + B rather than the Williams et al., (1988) relation log g = A log m + B. But
Cheung et al., (1997) did not fit their data to a mathematical relationship.
Williams et al.,’s (1988) model has a thickness d of the outer
spherical shell open to several interpretations. These authors suggested
that d is the diffusive penetration distance of the agent. However, since
the diffusion distance increases with the square root of time (Vt), it should
change as the tumor grows; it may not grow as (V t) exactly. The Williams et
al., (1988) diffusion length interpretation was experimentally deduced to
be a constant proportional to log(d). Endrich etal., (1979) noted that well-
vascularized outer layer of the tumor is 1 0 0 -2 0 0 gm thick while the semi-
necrotic layer below this, less well perfused, measured 500 - 1000 gm. d
could also be the thickness of the interstitial pressure boundary layer
predicted by Baxter and Jain (1989), to wit ( - v l(K/(LpS/V )), where K is the
hydraulic conductivity, Lp is the vessel hydraulic conductivity, and S /V is
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
51
the vascular density. These authors estimate d to be about 0.03 cm. The
interstitial pressure is so nearly equilibrated with vascular pressure that
almost no convective drug transport can occur from blood to tumor.
The same power-law was found by Williams et al., (1988) for binding
and non-binding antibodies relative to tumor size. A fourth interpretation
of d could be ruled out as resulting from a “binding site barrier”
(Weinstein and van Osdol, 1992). But this interpretation is supported by
Beatty et al.’s (1989) study which showed that predosing with large amount
of unlabeled antibody led to the subsequent uptake of labeled antibody
independent of tumor size. The saturation of the binding sites near the
tumor periphery thus allowed the drug to penetrate deeper into the tumor.
Blumenthal et al., (1991)increased the total dose of antibody to saturate
peripheral binding sites, and this produced a more homogeneous
distribution only in some tumors. Variability between tumors may account
for the difference between the results Williams etal., (1988) and Beatty et
al. (1989).
The mathematical models introduced by Baxter and Jain (1989) in a
series of influential publications dealing with a quantitative cognition of
the obstacles encountered in intravenous drug delivery to tumors, showed
that absence of lymphatics in tumors, in association with high tumor
vascular hydraulic conductivity, would lead to high interstitial pressure in
the interior of the tumor, decreasing to the level of the surrounding tissue
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
52
at the periphery in a thin boundary. Such high interior pressure
produced two effects: A greatly reduced transmural pressure difference
for interior vessels (1) which reduced greatly convective drug transport,
and (2) a convective current directed out of the tumor. The model of
Baxter and Jain (1989) shows how for idealized spherical tumors tumor size
influences these effects.
These authors provided the first spatially distributed models for
tumors (Baxter and Jain, 1989), a valuable tool for determining the
limitations of the well-mixed compartment assumption for tumors.
However, these models considered only tumor and neighboring normal
tissue, and does not provide a whole body framework to assess the delivery
to the tumor relative to that in other organs.
The Baxter and Jain model provides a tool for the reexamination of
interpretations of the literature data on drug distribution. Ahmad etal.,
(1993) for example, reported success with iv-injected immunoliposomes
encapsulating doxorubicin directed at squamous lung carcinoma. They
determined that the periphery of the tumors was the binding location in
the main of the antibody, and they suggested that this was selective
binding to the highly proliferative peripheral stem cell layer. The pattern
of such binding is not unique to this antibody, and is quite likely to be due
to poor perfusion in the tumor interior combined with other obstacles as
shown in Baxter and Jain’s model.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
53
The prediction of a convective flux directed to the exterior from the
tumor center suggests that the diffusion times may underestimate the time
for a drug to reach areas distant from vessels was shown by Baxter and
Jain.
Hybrid models have been developed by taking into account spatial
variations within tumors, but other organs are still treated as well-mixed
compartments. Terasaki et a l (1985) provided a mathematical model to test
the validity of the assumption of well-mixed compartments for non
eliminating organs. They concluded that a well-mixed assumption adequate
for adriamycin and several other drugs. The model accounts for the
binding to the tissue and the extravasation of the drug leading to a plasma
concentration decrease along the vessel. The assumption is made that the
drug distributes itself uniformly throughout the tissue after extravasation.
Thus, Teresaki et al.’s (1985) model is unconvincing for assuming that al
non-eliminating, non-tumor organs are well mixed. So, further work is
needed.
During the passage from the arterial to the venous circulation mass
transfer into tissue will obviously decrease the plasma concentration. If
the fraction of the transferred drug on a single pass is small compared
with the total injected amount, then well-mixing may be assumed. The
advantage potentially, of intra-arterial delivery feeding directly to the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
54
tumor-bearing organ, is that a significant amount of drug is transferred
on the first pass and on the significant clearance of the drug eliminating
organs on the first pass. Some models allow for this, as seen in Table 1.6, by
replacing the singly well-mixed plasma compartment with different
plasma concentrations in different organs.
Table 1.6 Various Mathematical Modeling Approaches to
Plasma Concentration of Drug.
Single plasma compartment
• Immunotoxins Mice T
• Bifunctional antibodies M ouse/general T
• Streptavidin-conjugated antibodies General T
• Biotinylated antibody Guinea-pig T
Different plasma concentrations for different organs
• Doxorubicin and others Human T
• IgG antibody and fragments Mice NT
• Monoclonal antibodies Mice T
• Monoclonal antibodies Mice, humanT
Note: T=tumor present; NT=no tumor; L=lymphatics included; NL=no
lymphatics.
The well-mixed assumption for non-eliminating organs investigated
by Terasaki et al., (1985) addressed the question whether plasma
NL
NL
L
L
NL
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
55
concentration decreased significantly in passage through the organ. They
simplified the time scales for drug binding and drug diffusion from the
vessel outwards into the tissue, assuming these were instantaneous, thus
the tissue concentration varied only in the direction of the vessels, and
then only because the plasma concentration varied. They concluded that
non-eliminating organs were well-mixed for doxorubicin (and some other
small drugs), suggesting that only a small fraction of the drug extravasates
on the first pass through these organs. The implication of Terasaki et a l.’ s
(1985) paper is that infusion of doxorubicin (and the other drugs
considered), intra-arterially into a non-eliminating organ would not
increase tum onnorm al delivery.
If the drug circulates first through the tumor bed one might exploit
a significant change in plasma concentration on the first pass through an
organ(s). For liver metastases of colon cancer (Rougier etal., 1996) this
rationale for intra-arterial delivery has been found superior to
intravenous delivery. However, intraarterial (ia) infusion has been
studied for livers (Simonetti et al., 1997), stomach and breast cancer
(Murakami et al., 1997) with disappointing results. Hassenbusch et al.,
(1996) said in a study of ia therapy for brain tumors, “Although these data
show higher drug concentrations with ia infusions, actual values were
considerably less than predicted by theoretical modeling. This discrepancy
between theoretical and experimental results emphasized the need for
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
56
further study of causes and remedies so that ia therapy can achieve better
drug concentrations with less toxicity”.
The interstitial pressure boundary layer is about 0.03 cm (Baxter
and jain 1989), while the well-vascularized layer on the outer edge of a
tumor (Endrich etal., 1979) (below which perfusion is much reduced) is
0.01 to 0.02 cm thick. These two length scales were estimated
independently. Their similarity suggests an association between high
interstitial pressure and decreased perfusion in tumors. Literature has not
brought forth this point, although a mechanism for vessel collapse was
proposed via high interstitial pressure.
Baish et al.’s (1997) model allows for recirculation within the tum or
between vessels of high and low pressure and takes into account the high
vascular conductivity of tumor vessels; this model being one for
distributed vascular and interstitial pressure. The results obtained from
their model were interpreted to explain the diversion of flow from the
tumor center to the periphery. In their model vessels in the center do not
actually collapse but they exhibit a “mechanical constriction” of central
vessels’ cross section. Zlotecki etal., (1993) utilizing angiontension-II to
induce hypertension, claimed that “parallel increases in tumor interstitial
pressure (TIFP) and tumor blood flow (TBF) did not support that elevated
TIFP causes vascular collapse and thus decreases TBF”, Jain (1994) arrived
at a similar conclusion by measuring pressure in superficial postcapillary
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
57
venules and discovered that it was quite close in general to tumor
interstitial blood flow, and that both varied little in the tumor.
Furthermore, Jain (1994) found that “the variation in microvascular
pressure between vessels of a given tumor was generally small. Large
recirculation effects or counter-current exchange is difficult to allow for
in Baish et al.’s (1997) model. However, Jain were only capable of
measuring superficial vessel pressures.
“A lack of correlation between alteration of mean arterial blood
pressure and changes in tumor perfusion” was noted by Stone e ta l.,(1992)
experimentally in mice. “Blood flow in tum or tissue increased selectively
when the blood pressure was elevated by infusion of angiotension II”
reported Sato etal. (1995); Jain (1994) found the principal driving force
for interstitial hypertension in solid tumors to be “microvascular
pressure”. Sato e ta l’s (1995) results are positive evidence consonant with
Stone et al.’s (1992) is a causative factor in low tumor perfusion. It is
possible however that different trends in tumor perfusion axe observed
depending on which location in the tumor is measured. Their model
(Baxter and Jain, 1989) claims that total perfusion increases as vascular
pressure is increased but only in the thin outer periphery of the tumor.
The ambiguous relation between interstitial pressure and tumor blood flow
needs a resolution.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
58
Useful models for understanding local heterogeneity in
vascularized tumors were considered by Baxter and Jain (1989) all with a
local Krogh-type approach for tissue surrounding a blood vessel in the
case of bifunctional antibodies and enzyme-conjugated antibodies as
considered by Baxter et al. Convection is neglected in both models by
spatially varying reaction and diffusion were taken into account. No
convection as an assumption is not explicitly justified in these
publications but it may correspond to a situation of a tumor interior where
interstitial pressure is almost equilibrated with vascular pressure, so that
filtration through vessel is severely limited. Himmelstein (1973) assumed
that plasma concentration was proportional to the tumor concentration.
Spatial variation was not used in this model unlike time dependence. These
assumptions may apply since methotrexate is used for leukemia.
The distribution between plasma and tumor concentration is
important for models of solid tumor cell kinetics under chemotherapy, like
amsacrine many anticancer drugs, antimitotic or antitopoisomerase
agents, acts only during a cell cycle’s critical phase, i.e., the S-phase. For
Lewis lung tumor in the mouse the half-life of amascrine is 2.7 h, much
higher than that of plasma, 0.3 h (Baguley and Finlay, 1995). The median
duration of this tumor is 9.6 h for 0.34 cm size and 12.4 h for 4.2 cm size
according to Steel (1977) (units not given). Median doubling was 5.3 h and
158 h, respectively. Thus a plasma drug concentration model for cell kill
would be in serious error.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
59
1.6 MAGNETIC RESONANCE IMAGING (MRI)
The following discussion is heavily indebted to a review by Singh
and Waluch (2000). In just over a decade magnetic resonance imaging
(MRI) has become the imaging modality of choice for the non-invasive
study of lesions and tumors in living organisms with broad application in
the abdomen, pelvis and musculoskeletal system. Concurrent development
of contrast media, now in widespread use, has aided the rapid expansion of
this field and increased clinical efficacy. MRI offers high spatial
resolution and soft tissue contrast, with sensitivity to contrast media
greater than that of x-ray computed tomography (CT). MRI provides
excellent soft tissue contrast on unenhanced images but contrast
enhancement improves substantially sensitivity and specificity of the
area examined.
1.6.1 GENERAL THEORY OF MRI
MRI can be thought of as measuring and mapping the magnetic
properties of an object. But instead of measuring the magnetism point to
point in the object, MRI can create an image that depends on the rates of
magnetization for each point. Different materials have different magnetic
properties, and will behave differently in an applied magnetic field B o.
Spinning charged particles produce a magnetic field directed
perpendicular to the plane in which the charged particles spin. Thus,
such spinning particles may be regarded as tiny bar magnets which will
line up with B q. Matter is made up of electrons, protons, and neutrons.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
60
Electrons and protons, as well as nuclei that have net angular momentum
have spin and are charged, and will line up in B o- For quantum
mechanical reasons, a proton for instance, can only line up parallel or
anti-parallel (north pole to north and south pole to south) with respect to
B o- The anti-parallel alignment is higher in energy than the parallel one,
analogous to the bar magnet. A hypothetical single proton in empty space
would find itself locked into one of these two positions, as a spontaneous
transition from the higher energy state to the lower is a very slow
process. To go from the low energy state to the higher one requires
energy, of which there is none in empty space. In real materials, for
example water, there is a myriad of protons moving about and scattering
off each other as well as all the neighboring molecules. Protons that find
themselves “locked” in the high energy state would soon find a willing
acceptor (lattice) that would absorb the transition energy and the proton
would drop to the low energy state i.e., aligned with Bo. Similarly, protons
in the low energy state would soon receive from the lattice the exact
energy needed to take them to the higher energy state. Thus, in real
materials there is a dynamic alignment-misalignment process that soon
reaches equilibrium and the energy states would be populated according to
the details of the Boltzman equation with a slight excess of protons in the
low energy state. Thus, a bulk magnetization is created in the material. It
should be noted that this bulk magnetization for most tissues is
exceedingly small and can only be measured by sophisticated techniques
such as MRI.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
61
We could radiate the sample with radiofrequency electromagnetic
waves (RF) whose energy would exactly match the transition energy
between the states and thus cause the protons in the lower energy states to
jump into the high energy states. We could thus modify the relative
population of these states, and thus change the bulk magnetization, so long
as we kept pumping the RF into the sample to population of states. Once
the RF pumping was discontinued, the high energy state protons in the
material would continue to dump their transition energy to the lattice and
repopulate the lower energy states until the Bolzman equilibrium was
reestablished. These transitions can be detected with an appropriate
antenna and thus a signal detected.
Looking at this process classically, the bulk magnetization is forced
to tilt away from alignment with Bo by the oscillating RF magnetic field
(we can ignore the electric component of this field), as shown in Figure
1.6. While the bulk magnetization is tipped away from alignment, it will
process around Bo for the same reason that a spinning gyroscope
processes around the direction of gravity (merely replace the force of
gravity with the force due to Bo , Figure 1.7). If a conducting coil
(antenna) is positioned appropriately with respect to this precession, an
oscillating current will be induced in the coil (Lenz’s law). The process of
establishing the Boltzman equilibrium is called relaxation.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Rotation of the Bulk Magnetization by 90° away from B 0 .
Figure 1.6
Precession
Precessing Proton
Figure 1.7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
63
1.6.2 Tx RELAXATION
In the process described above, the high energy state protons give
up their transition energy to a lattice that is willing to accept that precise
amount of energy. It is the willingness of the lattice to accept these
energies that determines the rate at which the high energy protons
repopulate the lower energy state. Conversely, it is the willingness of the
lattice to provide this exact quantum of energy that determines the rate of
low energy state protons transitioning to the higher energy state. This
“willingness” rate characteristic is a property only of the lattice and not
the proton. It is medically fortuitous that different tissues, both normal
and diseased, have vastly different willingness rates. The recovery of this
magnetization can be described by a single exponential equation with a
time constant called Ti, and the process is termed Ti relaxation. This is
illustrated in Figure 1.8.
Short T 1
LonflTI
Tit Time
t, Relaxation.
Figure 1.8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
64
1.6.3 T2 RELAXATION
This relaxation phenomenon is best understood ciassically. Take the
situation where the magnetization of a given voxel has just undergone a
tilt away from Bo and is processing. We can mentally subdivide this voxel
into a bunch of smaller voxels each of which has a processing magnetic
moment and these moments process in sync and alignment so as to add up
to the total magnetization of the parent voxel. So long as this coherence is
maintained, all we will see is a Ti relaxation phenomenon. However, if the
precession rates in these smaller voxels differed even by tiny amounts,
then soon the small magnetization vectors would start pointing every
which way and start canceling each other, and the total amount of
magnetization processing (called the transverse magnetization) in the
parent voxel would rapidly decline. What we would see in our antenna is a
rapid decline of the signal. This decline is monoexponential and is
described by time constant T2. The original of this dephasing is due to a
variety of phenomena including the local randomly fluctuating magnetic
fields surrounding groups of protons that cause random changes in the
processional frequencies, to the ability of neighboring protons to
exchange the transitional energy and thus move the magnetization state to
new areas where the processional frequency may be different, as well as
to the nonhomogeneity of the applied magnetic field B q. Note that in this
relaxation no energy is exchanged with the lattice, so that T2 relaxation is
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
65
entirely independent of Ti relaxation. It is also medically fortuitous that
tissues have very different T 2 values (Figure 1.9).
M r ,
Signal
LongT2
S h ort T2
0
Tima TB
T2 Relaxation. Loss of Precessional Coherence.
Figure 1.9
1.6.4 FORMING IMAGES
From the above discussion it is clear how a signal can be generated
by a sample and deleted. It is also clear that the strength of the signal
detected will be different depending on when we interrogate the
magnetization during its various relaxation. We can generate a signal that
is based only on Ti by detecting the signal just after the RF irradiation, or
we can mix in T2 dependence by waiting several milliseconds after the
irradiation, and letting T2 relaxation happen.
The question remains, how does one map these voxels into an image,
i.e., where in space these voxels are that gives us the signals? If we are
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
66
dealing with a three-dimensional object, as we usually are, the first
problem is to limit the number of voxels with which we are dealing. This
can be accomplished by using localized excitation. If we were to add a
small field gradient parallel to Bo (assume to point along a Z-axis
coordinate), then the protons at different points along the Z-axis would be
resonant at slightly different frequencies. We could thus excite a plane of
these protons by choosing the appropriate excitation frequency. Once
excited, however, all the protons in the plane would send back signals at
the same frequency, since they are all located at the same net magnetic
field. So we need another trick. The simplest trick is frequency encoding.
Just before the protons start radiating their transitional energies, we
could apply another magnetic gradient that was in the plane of excitation,
say along the X-axis. Each column of protons (perpendicular to the X-axis)
would find itself in a slightly different magnetic field as compared to its
neighboring column. All the protons within each column would then
radiate energies according to their new processional frequency. Since we
know that strength of the applied magnetic gradient exactly, we could
form a one-dimensional (ID) map of the magnetization of dimensional
maps are rarely useful. To encode information in the other dimension (so
that a two-dimensional (2D) image can be created) we could resort to
repeating this experiment but each time rotate the axis for frequency
encoding several degrees until we went around a full circle. Then we
could resort to a back projection reconstruction algorithm, as this set of
signals would be analogous to the information obtained by a CT scanner.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
67
Alternatively, a more sophisticated idea can be used. If just prior to the
application of the frequency encoding readout we were to apply a short
duration burst of a small magnetic field gradient perpendicular to the
frequency readout direction, we could modulate the received signal a little
by causing a change in the processional frequencies along this gradient,
and thus a signal loss due to the resulting dephasing of spins along this
gradient direction (that is in each column). If we now did this n times and
each time increased the strength of the gradient by a constant small
amount, we would create a periodically encoded data set in this second
dimension, and therefore, we would have created a data set that was
encoded in two perpendicular directions. These who are familiar with
Fourier transforms can recognize that a two-dimensional Fourier
transform of this data set will yield a two-dimensional (n X m) image of the
tissue magnetization.
By applying the slice section gradient magnetization along an
arbitrary direction, an image can be created along this direction, so that
we are no longer confined to axial cross sectional images is in CT.
Depending at which time (in terms of Ti and Ti relaxation) the
magnetization is interrogated or how long the tissue is allowed to relax,
images can be created whose contrast differences can be based on the
degree of Ti or T2 relaxation, or complex combinations of these relaxation
processes. The RF pulse can be designed to tilt the magnetization a fixed
number of degrees from B q, and this can also vary the contrast in the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
68
image. Images of flow can be obtained, obviating angiography in many
cases. By further extending the above principles, three-dimensional (3D)
images can readily be obtained. Each year new and exciting clinical
sequences are developed that add to the clinical armamentarium.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
69
2. SPECIFIC AIMS
1) To design and manufacture surface magnetic resonance
spectroscopy coils of various sizes for fluorinated drug studies in
vivo which would be suitable for patient use and give the best
signal-to-noise ratio (S/N ratio), sensitivity, and patient comfort.
19
2) To use these surface coils for in vivo F-magnetic resonance
spectroscopy to distinguish between patients whose tumors trap 5FU
and those who do not, and hereby confirm the work of oncologists
in our team.
3) To develop a non-invasive, in vivo imaging technique employing a
contrast agent which quantifies tumor vascularization, and possibly
tumor osmotic pressure efficiently and more rapidly than is
possible by present 5FU techniques, with reduced discomfort to the
patient.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
70
SECTION 3 (SURFACE COIL)
3.1 SURFACE COIL DESIGN AND PRACTICE
In medical magnetic resonance imaging auxiliary coils may be used
for location on the surface of a patient, and acquire information of
specific portions of tissue, organs or fluids non-invasively. These coils
may be connected to existing whole body magnets and their supporting
electronics and computer hard- and software.
There are a number of coils of different design, some single
resonance or multiple resonance. Our work will concern simple
resonance coils designed and manufactured in-house. These coils will be
called STL coils, for Surface Trans-Locatable coils which are intended to
achieve better sensitivity, enhanced signal-to-noise (S/N) ratio, locatable
and fittable to the patient.
3.1.1 INTRODUCTION
Our basic STL copper coil design consists of a circular loop of wire
composed of one or more turns, wound either in a cylindrical or planar
fashion. The coil is tuned to the NMR frequency of interest and matched to
the impedance of the driving device, typically 500. One standard tuning
and matching scheme is the capacitively-coupled resonant tank circuit
shown in Figure 3.1. Degradation in circuit performance resulting from
electrical coupling to loss of signal from conductive samples tissue, fluids,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
71
organs, etc. can be reduced by balancing the NMR coil with respect to
ground (Decorps et al, 1985; Nagel et al, 1990). An example of a
capacitively-coupled balancing scheme is shown in Figure 3.1.
Double Tuned STL Surface Coil and 50D Coaxial Line
Figure 3.1
(a) Capacitive tuning and matching, viewed as an imaginary voltage
divider network
(b) Symmetric matching capacities uncouple the coil from feed ground
and reduce capacitively induced sample losses
This single-resonance STL coil probe has been extended to
applications in multi-nuclear NMR spectroscopy through development of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
72
multiple-frequency circuit-design technology(Cross et al, 1976; Kan et al,
1980; Doty et al, 1981, Schnall et al, 1986; Reo et al, 1984; Grist et al, 1988;
Eleff et al, 1988). The use of resonant transmission lines and lumped
reactive elements in multiply-tuned circuits is discussed in detail in the
literature.
Development of the circular surface-coil design have been
motivated by a need to accommodate better the sample size or shape, to
enhance the signal localization, and to improve the sensitivity of the coil.
NMR coils balanced with respect to ground (through capacitive
coupling or inductive coupling(Decorps e ta 1 , 1985) have been shown to be
effective in reducing dielectric losses in conductive samples like tumors.
Dielectric losses have been reduced also by symmetrically distributing
lumped capacitors around the coil to reduce the potential difference
between the coil and ground(Schenck et al, 1986; Decorps et al, 1985).
Explicit grounding at the coil center and electric shielding of the sample
from the coil are features of the "crossover" surface coil which minimize
dielectric losses in the sample(Nagel et al, 1990).
STL coil as belonging to an important class of coils used for local
signal detection is based on the loop-gap resonator design(Grist and Hyde,
1985). These low-loss, high-sensitivity coils consist of wide conductor
bands that are brought to resonance with capacitive gaps. Two coplanar
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
73
loops that are laterally displaced and connected by a single gap from the
planar-pair loop-gap resonator(Hyde et al, 1986). This coil detects signals
from a slab-shaped region close to the coil and is useful for imaging
superficial structures such as the temporomandibular joint. A further
advantage of this design when used as an imaging coil is that the two loops
are intrinsically isolated from an external homogeneous field of arbitrary
orientation, such as a homogeneous transmitter field. This results from
the fact that, for the proper circuit resonance condition, an emf induced at
the gap by flux linking one of the loops will cancel that induced by flux
linking the other loops. The counter-rotating-current coil consists of two
coaxial loops that are axially displaced and support current flows in
opposite directions(Froncisz et al, 1986). It, too, is intrinsically isolated
from an external field. Combinations of the planar-pair and counter-
rotating-current coils produce a resonant structure whose elements are
intrinsically isolated both from each other and from an external
transmitter field(Hyde et al, 1987). Additionally, the signals induced in the
two coil components can be detected in quadrature and combined to
produced a net V2 improvement in the signal-to-noise ratio (Chen et al,
1983).
3.1.2 SHIELDING
The purpose of shielding of the STL coil is to prevent the NMR
resonator from behaving as an antenna, by partially or completely
enclosing it in a more or less perfectly conduction or Faraday "cage", with
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
74
the aim to confine its field energy to some finite volume, and to prevent
stray external signals from entering. This prevents radiation losses,
sensitivity to outside noise sources (such as radio broadcast stations) and
interaction with lossy or non-ideal outside materials such as tissue, fluids,
organs, etc. A perfect shield acts as a "mirror", such that all currents
flowing in the antenna create oppositely flowing currents in the shield.
Since the total external flux created by the currents flowing in the
antenna structure thus cancels to zero (more precisely: the external dipole
characteristics are transformed into quadrupolar characteristics with
correspondingly steeper field decay), there is no effective field and no
energy stored outside the shield. This effect is directly comparable to the
working principle of commercial gradient shielding which induced
electromagnetic noise. Well-designed rf shielding structures allow high-
frequency currents to flow at minimal resistance, while low-frequency
currents, such as those induced by switching gradients also known as eddy
currents, should find a high-resistance path. Otherwise the probe interior
would be shielded from the gradient fields as well. Therefore, if gradient
fields have to penetrate the body, it is recommended, not to use plain
copper foil, but coarse copper mesh structures that does not offer DC-
current loops but capacitively connected loops only.
3.1.3 COUPLING
The coupling mechanism to the MRI unit is responsible for feeding
transmitter power to the probe to replace continuously the energy lost in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
75
the sample and probe during a pulse, and to transfer signal energy picked
up by the probe to the receiver. Energy can be transferred to the probe
where it provides sensitivity. It can be coupled electrically (capacitively)
to positions where the probe's internal electric fields are high, whether as
combined elements or singly in the probe structure). Alternatively,
energy can be transferred by magnetic fields e.g., inductively at positions
of high intrinsic magnetic field contained in the coil.
Capacitive coupling to a single current loop such as a surface coil is
illustrated in (Figure 3.1). The driving voltage is applied to the probe's
capacitor through an adjustable matching capacitor. The external
matching and the internal tuning capacitor act as an imaginary "voltage
divider", where Ct determines the resonance frequency, and the ratio Ct~
VCCfi+Cnr1) determines the partial driving voltage and the coupling
strength. More rigorous concepts of coupling as an impedance matching
processes have been developed and can be found in the literature(Hoult,
1979; Decorps et al, 1985).
It must be emphasized that tuning and matching are basically
independent functions. In many cases, the "voltage divider" is set up with
a variable matching capacitor which is connected to a fixed capacitor of
the probe, while probe tuning is achieved by one or more variable
capacities elsewhere in the probe.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
76
In the coupling scheme of (Figure 3.1) one end of the coil is kept to
feed ground, while the other end assumes maximum electric potential. The
loss effect due to this coil potential can be strongly reduced if a balanced
feeding scheme is employed (Figure 3.1). In this symmetrical scheme the
coil is uncoupled from ground and assumes a virtual referencing ground
potential to the MR equipment in its center, such that the electric
potentials at both coil ends and the corresponding energy losses are
reduced by a factor of two(Decorps et al, 1985). Virtual ground positions
can easily be detected by touching with a finger tip the coil connected to a
sweep generator. The higher the electric potential, the more pronounced
is the shift and widening of the coil resonance in the spectrum. Touching
the coil at a virtual ground position barely affects the appearance of the
resonance.
Inductive coupling is less frequently used in combination with
surface coils, but it is instructive to demonstrate the principle with this
simple example (Figure 3.2).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
77
Weak coupling
Strong coupling
•o
Frequency
STL Coil Coupling Profile
Figure 3.2
The primary surface coil resonator, tuned to the desired resonance,
is approached by a coupling surface coil resonating in the vicinity of the
main coil resonance. The primary coil is not connected to ground;
therefore inductive coupling provides intrinsic balanced feeding. At some
distance, the sensitive volumes of both coils barely overlap, and
practically no energy is transferred from the driving coil to the main
resonator (Figure 3.2). If the coupling coil approaches the main coil, the
magnetic flux common to both coils is increased and the coupling becomes
stronger. This behavior can be monitored if the coupling loop is
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
78
connected to a sweep generator. Closer coupling corresponds to less power
reflected from the coupling loop at the resonance frequency of the
primary circuit, as the power is absorbed by the primary loop. Perfect
match is achieved when all power is absorbed by the primary coil, and no
power is reflected into the transmitter at the desired resonance frequency.
Alternatively, coupling strength can be increased by bringing the
resonance of the coupling coil closer to the main coil's resonance
frequency (Figure 3.2), again increasing the flux from the coupling coil to
the main coil in the desired frequency range.
Coupling strength can be increased by spatially approaching the
main resonator with the coupling coil, to increase the flux common to both
coil, or alternatively by tuning the coupling coils resonance frequency
towards the main coils resonance to increase the coupling coils' energy
content at the desired frequency.
In both cases of capacitive or inductive coupling, the coupling
device can be viewed as a second resonating structure overlaid to, and
providing the energy loss by, the main resonator. With respect to field
homogeneity, this has no consequence if the symmetry of the coupling
device matches the main coil symmetry. Probes that can be characterized
by a single current loop (no parallel paths in the magnetic field-
producing conductors) are best coupled capacitively, leaving the
symmetry of the current distribution in the inductors unaltered.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
79
Inductive coupling by a surface coil-type mechanism inevitably overlays
the surface coil field characteristics onto the main coil field distribution.
In low-loss probe/sample combinations only weak coupling is necessary;
the coupling coil provides only a small percentage of the magnetic field
energy stored in the probe. High loss combinations require high energy
flow through the coupling device, and thus result in higher
inhomogeneity of the rf power.
3.1.4 L-C RATIO
There exists an unlimited number of possibilities for the
construction of a 3 cm 1 9 F surface coil at 1.5 T: A single turn thick cooper
wire coil in combination with a fairly high capacitance of more than 30 pF
will lead to the desired resonance at 59.8 MHz, as good as a 2 turn copper
wire coil with a correspondingly lower capacitance of approximately 1 0
pF. Other relatively inert metal may be used instead of the cost-effective
copper.
Two criteria lead to a narrower range of possibilities: In principle,
one wants the lowest possible inductivity of the coil in order to achieve
high currents at low voltages corresponding to high magnetic fields at low
electric fields. This requires high capacitance values, which are also
favorable with respect to detuning due to changing the sample or patient
movements, On the other hand, one wants to minimize losses in the coil's
connection leads which cannot be arbitrarily shortened. If the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
80
connection leads length is 1/3 of the total length of the coil wire, roughly
1/3 of the magnetic energy does not contribute to the desired sample
interaction, but is dissipated in the coil. Therefore one has to optimize
towards short and thick probe wires, that are still substantially longer
than the connection leads. It must be strongly emphasized that long
connection leads (relative to the total coil wire length) lead to serious
degradation of probe efficiency.
3.1.5 SENSITIVITY VS. HOMOGENEITY
A good resonator is characterized by both high sensitivity and high
field homogeneity. Homogeneity is optimized by evenly distributed
currents arranged with high symmetry. Examples are the “solenoid” and
the “birdcage” resonator, with high longitudinal and pronounced
rotational symmetry. The relationship between symmetry classes and field
homogeneity properties for all kinds of magnetic field generating devices
(applicable to the design of main magnets as well as gradient coils) was
discussed in general by Romeo and Hoult (Romeo and Hoult, 1984).
Sensitivity is optimized for a given resonator and a given sample
(assuming all electric loss factors have been eliminated) if the sensitive
volume of the resonator is identical to the dimensions of the sample (or the
region-of-interest to be measured).
If a sample of cylindrical dimensions, such as a solution in an NMR
tube, is to be spectroscopically analyzed, the use of a matching solenoid
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
81
results in almost perfect homogeneity across the sample at simultaneously
maximum sensitivity. Conflicts between sensitivity and homogeneity
arise, if the sample is much larger than the desired region of interest, e.g.,
if a transverse thin slice image of the sample is to be acquired. A much
longer solenoid than the desired slice thickness shows good field
homogeneity across the slice, but it will pick up noise from parts of the
sample not contributing to the signal and degrade sensitivity. If, on the
other hand, the solenoid is shortened to the desired slice thickness, its
sensitivity is strongly enhanced at the cost of decreased homogeneity.
Bottomley etal., (1988) have shown that the sensitivity of a birdcage
resonator is increased by 2.6, if instead of the length/radius of 4, which
yields optimum homogeneity, a length/radius ratio of 1.4 is used. This
decrease in homogeneity is less severe for imaging of transversal slices or
spectroscopic voxel localization than for longitudinal slices. Hoult(Hoult,
1979) showed that the transformation of a two turn surface coil into a
Helmholtz coil decreases sensitivity by 13.4% with a corresponding gain in
longitudinal field homogeneity. In both cases this change in sensitivity is
directly related to the change in the sensitive volume of the coil. It is
therefore recommended to design resonator geometry specific to the
desired samples, i.e., with sensitive/homogeneous regions not extending
substantially outside the typically desired field-of-view dimensions. At
this point one has to consider if employing a large transmitter probe with
homogeneous transmitting fields and a separate small reception probe
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
82
yielding optimal sensitivity would be preferable. This choice, however,
calls for measures of active or passive probe decoupling (Hyde and
Kneeland, 1988). Active decoupling must be effected to the receive coil
during transmission of the signal to avoid burning out the receiver
network. This is done by means of system hardware. With passive
decoupling one mechanically detunes the receive coil.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
83
3.2 SURFACE COIL EXPERIMENT
3.2.1 COIL REQUIREMENTS
19
The essential steps for obtaining values of F metabolite
concentrations are:
19
1. To transmit and receive at the F resonance frequency and, able to
pick up a 1 H frequency signal for imaging,
2. Total depth of signal penetration to be selective via coil size,
3 . To determine the relative position and orientation of the chosen VOI (as
used in a clinical MRS study) within a suitable sensitivity map of the coil,
4. To integrate the sensitivity over this volume,
5. To use this information to scale the spectral peak
amplitudes, together with factors to account for coil
loading. The measured signal from a reference sample
in a fixed position may then be used to convert relative
peak amplitudes into metabolite concentrations.
6 . To safeguard against excessive rf radiation of patients.
3.2.2 CIRCUIT DESCRIPTION
The double resonant design (Figure 3.1), is adapted from a single
frequency coil that makes use of semi-rigid coaxial line segments to
distribute capacitance symmetrically about the coil, thereby reducing its
peak rf electric "E" field by establishing a bilateral "E" field balance about
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
84
a virtual ground "0 " central to and aligned with the lead axis of the coil.
This ground potential of the coaxial shield eliminates coil tuning and
match instabilities resulting from sheath currents.
The centered ground minimizes "E" field coupling to the sample as
well. Lower peak voltages in the coil circuit result in an improved peak
power rating for this design. Viewed as two integrated loops, this coil
design is inherently double resonant. The high frequency loop is
composed of the solid conductor "c" capacitively coupled with the coaxial
line shields which in turn are connected at "d" to complete the circle. This
capacitive splitting of the high frequency loop serves to increase the
frequencies and diameters achievable, as well as to distribute more evenly
the short wavelength rf current for improved Bi field homogeneity.
Variable capacitors "a" and "b" provide for tuning and matching of the
high frequency mode, and for electrical balance adjustment for the entire
structure. The low frequency resonance is established by the more
inductive loop consisting of the solid conductor connected to the center
conductors of the coaxial line elements of the coil. Tuning, matching, and
balance of the low frequency is performed by the more conventional
circuit shown. Although capacitance adjustments to the two loops are
somewhat interactive, two resonance’s can be optimized independently for
mutual tuning, matching, and balancing over a range of load conditions,
This coil can be divided into more segments to increase further diameter
and/or frequency.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
85
EQUIPM ENT USED
In-vivo NMR 1.5T Magnetom
Siemens Medical Systems Inc. Iselin, NJ 08830
Network/Spectrum Analyzer (HP 4396A)
Hewlett-Packard Co.
5161 Lankershim Blvd. No. Hollywood, CA 91609
Signal Reflector Circuit
ANZAC Adams-Russell Co. Inc. Burlington, MA 01803
PARTS USED
Copper tubing O.D.=0.5 cm
Ceramic Capacitors (fixed) (non-ferromagnetic)
RF/Microwave Capacitors
Values used from lpF to 600pF
American Tech. Ceramics Corp. Huntington Station, NY
Ceramic Capacitors (variable) (non-ferromagnetic)
RF/Microwave Capacitors
Values used lpF-15pF (variable)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
86
lpF-30pF (variable)
Johnson Mfg. Corp. Boonton, NJ
RF Cables (50 ohms) (non-ferromagnetic)
Alpha Wire Corp. Elizabeth, NJ
3.2.3 COIL TESTING (BENCH WORK)
Set the spectrum analyzer to which the STL coil is connected to
center frequency of 63.88 MHz with a range of ±20.0 MHz, a HP 4396A
(Network/Spectrum Analyzer) was used to tune and match the surface coil.
A refractance circuit was set up by means of a 1.0 liter NaCl 9.0% doped
solution used as a load to measure the Q_of the coil. All output was printed
with a HP bubble jet printer (Figure 3.3)-
T r / I '
- 1 -
'
_ U /.J _
i
_ I : i
L ~ : ;
;
i
; •
I
:
1__ L L _ J _
DtV DIV CENTER S3 S0S 0 0 0 . Q00 H .»
3 cm Coil Profile with Load at Center Freq. at 63.9 MHz
Figure 3.3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
87
3.2.4 COIL TESTING (in vivo)
Experiments were performed with a loading phantom (1000 mL
saline solution contained in a plastic bag to simulate body loading to
ensure that there was no change in the sensitivity profile of this coil
when loaded. When applying this technique to other coil designs, it would
be prudent to ensure that there is no effect on the coil sensitivity profile.
This method assumes a uniform distribution of metabolites within the
volume of interest (VOI) and does not explicitly include correction for
losses during the localization process.
All experiments were performed in a Siemens 1.5 T Magnetom
system. The coil map and spectroscopy measurements were acquired with
19 l
a 3 cm F/ H transmit/receive dual frequency transmission line resonator
(TLR) surface coil(Zabel etal, 1987). A proton marker ring containing
water doped with NaCl and CUSO 4 to give a Ti of approximately 260 ms was
attached to determine the precise coil position in all images, because an
essential step in our quantification process is an accurate geometrical
location. The pulse used for all our measurements was a single 10 ms
adiabatic rapid half-passage excitation pulse(Bendall and Pegg, 1986) with
a bandwidth of 2.5 kHz. This ensured that the excitation was unaffected by
the spatial variation of the Bi produced by the surface coil and therefore
essentially independent of position. Also, this pulse was used to obtain the
clinical data for which this technique developed.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
88
3.2.5 COIL MAPPING
A coil map was obtained to determine the sensitivity variation of the
coil with a position assuming uniform excitation of spins. For this reason
the same large phantom was used to produce an identical solution to that in
the proton marker ring. An annular tube filled with distilled water
superposed on the coil to provide loop of the coil location in the image.
This phantom had to be large enough to extend beyond the extent of the
surface coil so that an accurate sensitivity coil map would be obtained. The
TLR was switched to 1 H mode (i.e., tuned for 1 H nuclei) with the concentric
proton marker ring attached (Figure 3.4).
Patient table
Experimental Arrangement for Acquisition of the Coil Map
Figure 3.4
The 3D STL coil map was obtained by means of a 3D FLASH gradient
echo sequence and the pulse described earlier. A 256 matrix size was used
Water _
phantom
TLR
Coil
Proton
marker
ring
Coil
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
89
with a repetition time of 2 . 0 sec. (which is in accord with the specification
that TR must be at least 5 times Ti relaxation). See General Introduction.
3.2.6 VOI DETERMINATION WITH RESPECT TO THE COIL
To correct for the sensitivity of a surface coil to a given VOI, it is
necessary to know their relative spatial positions and angles . For
localized spectroscopy, the VOI position is usually defined on a series of
images relative to the coordinate system of the magnet. To identify the coil
position in both patient images and coil sensitivity maps, a "proton marker
ring" is rigidly mounted on the RF coil. In MR images this appears as a
pair of dots whose separation varies according to where the image plane
intersects the ring. The position of the proton marker ring is determined
for both the patient images and the coil map data set using a cursor to
define the approximate position of both of the marker ring dots in each
image slice. The program then performs a center of gravity calculation to
determine the precise center of the proton marker ring dot and goes on to
calculate the coordinates of the center of the proton marker ring, the
angle of the coil with respect to the z axis and the offset in the z direction.
We utilized surface coils built to rigid specifications in our
laboratory. Instrumental sensitivity (low signal to noise ratio) presents a
problem in the case of 5FU studies in patients due to its low LD50. So, our
coils were built and tuned, to achieve highest S/N ration for each
particular instrument. A series of coils with different diameters (3 cm to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
90
15 cm) were used The coil may be placed close to the tumor site. The size of
the tumor determines the volume fields of the examination site. When a
small coil is used the acquisition volume is small and may not reach the
tumor site. A coil which is too large relative to the tumor size and location
will acquire not only the tumor but its surrounding environment. Thus,
our experiments aimed at utilizing optimum coil size. As an aside, for skin
melanoma a 3 cm coil was used, and at the other extreme a 15 cm coil was
used for tumors located 7 to 10 cm deep from the application surface. These
coils were tunable to each patient’s loading to increase S/N ratio. Loading
refers to an optimum radio wave energy transfer (shim coil and patient
inhomogeneity).
When a patient is placed in a magnet, the magnetic field of the
entire section has a magnetic inductive effect due to ferro-magnetic
materials and paramagnetic materials in the body which increases
magnetic inhomogeneity. To overcome this problem we shimmed the
applied magnetic field. Shimming in this local area means energizing and
de-energizing each of the subsidiary coils called shim coils, specifically
installed for this purpose, to achieve maximization of homogeneity of the
magnetic field (under absolute value and respiratory motion). The area on
which we focus (the liver) is not static. Respiratory motion imparts a
motion to the tumor causing it to change positions during data acquisition.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
91
3.3 In vivo STUDIES
Patients are placed in a MRI unit and imaged to verify the proper
positioning of the tumor to be studied in relation to the custom-designed,
19
say, 3 cm STL F surface coil. A background spectrum is acquired to verify
the absence of any prior 1 9 F signals.
An external reference standard of 1,2-difluorobenzene (DFB), placed
inside the surface coil, is used for chemical shift verification and for
referencing quantification.
A specified dose of 5FU is administered intravenously (iv) to the
patient, and serial spectra are acquired over 4.7 min. periods using both
localized chemical shift imaging (CSI) and unlocalized (global) sequence.
For the global sequence, a total of 256 free induction decays (FIDs) are
collected in each acquisition period with a repetition time (TR) of 1 sec, a
pulse width of ±2000 Hz, a vector size of 512, and an adiabatic half-passage
radiofrequency pulse to minimize differences of detection due to distance
from the surface coil to volume of interest(Wolf et al, 1995).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
92
3.4 RESULTS AND DISCUSSION
Commercially available coils at the time of our investigation were
designed for use by a wide variety of MR instruments. This imposed
limitations to maximization of the signal and the S/N ratio to any specific
MR instrument. Coil size was approximately 12 cm, and thus incapable of
focusing on small tumors. In addition while the coil might acquire signals
from tumor volumes in the 1 2 cm diameter range the unwanted signals
from adjacent areas between the tumors and depth of field obscured the
data from the actual tumors by adding extraneous signals. In terms of
tumor size and depth, the constraints for the commercially available fixed
size coil are as follows. A large tumor requires a large coil for data
acquisition; a deep seated small tumor requires a large coil; a deep-seated
large tumor also requires a large coil; a small tumor near the surface
yields unwanted signals with the large coil.
The localized placing of the commercial coil proved at times
impractical because positioning of the patient was restricted by the fact
that this coil had to be placed in one position, mainly on the scanning
table. In other words, the patient position had to be adapted to the coil.
Thus, sometimes, it was not possible to optimize the study, e.g., sometimes
the patient had to be fitted to the coil sideways, or abdominally, or even in
an oblique position. For the time span (often exceeding 2 hours) of the
total study the subject needed to remain in one exact position without
moving. Thus, the patient was stressed to limits often beyond endurance.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
93
19
In addition, no F coil was available and the commercial proton coil had to
19 19
be adapted for an F study. F signal optimization in these cases was very
difficult.
The advantages of the in-house manufactured surface coils were as
follows:
1. The coils were designed to meet the stringent electronic
requirements and peculiarities of our specific MRI instruments.
2. Individual STL coils were manufactured to be optimal for different
tumor size and depth. A series of coils were built to rage from 3 to
15 cm in diameter. This reduced the presence of stray and unwanted
signals when applied to a specific tumor area.
3. The placement of the STL coils was adaptable to the location of the
tumor in the patient, i.e., they could be placed anywhere on the
body of the patient to insure the subject’s maximum comfort.
Patient stress and pain was reduced significantly, even
dramatically, because we adapted the instrument to the patient
rather than vice versa.
19
4. Since the coil was specifically designed for F work the S/N ratio
was considerably enhanced.
5. These STL coils were cost-effective, and could be redesigned and
manufactured at will.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
94
SECTION 4 (MR SPECTROSCOPY)
4 1 9 F NMR SPECTROSCOPY
The principles of magnetic resonance experimentation and imaging
has been discussed in the General Introduction and in Section 3.
4.1 MAGNETIC RESONANCE SPECTROSCOPY (MRS)
The physical foundations of spectroscopy are the same as that
discussed for MRI. The primary difference is that in MRI magnetic
gradients are used to perform the magnetization acquisition signal; in
spectroscopy all such gradients are avoided. Another important
difference is that in MRI one can tolerate a quite spatially inhomogeneous
applied external field (Bo), and still create good images. In spectroscopy,
on the other hand, the Bo field over the region studied has to be extremely
homogeneous.
4.1.1 NUCLEI
Not all nuclei possess magnetic moments. Nuclei with spin 1/2 are
spherically symmetrical in magnetic moment. Of those that do, possess
magnetic moment multiples the spin quantum number may be greater
than 1/2 by integral. These nuclei can have nuclear quadrupole moments
as a result of their electric charge density deviating from simple spherical
symmetry, i.e., by having oblate or prolate spheroid symmetry. These
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
95
nuclei can interact with local electric fields and relax from excited states
via pathways more complex and rapid than those of spin 1/2 systems. The
study of such nuclei is called magnetic quadrupole spectroscopy
4.1.2 MR SIGNALS
The factors that determine the strength of an MRS signal depend
primarily on the number of the same nuclei present under the magnetic
coil, and on the energy between their ground and excited states. This
energy is related to the strength of the applied field Bo and the magnetic
moment of the nucleus.
4.1.3 CHEMICAL SHIFT
In the case of a perfect Bo field, one would expect all the protons in
an organic molecule to have the same resonant frequency, i.e. a single
peak in the frequency spectrum. However, this is often not the case. As
discussed in the MRI section, each proton frequency axis reflects the local
magnetic field related to Bo and modified by the magnetic moments of
neighboring nuclei by spin coupling as well as by the distribution of
electrons in the adjacent chemical bonds. Thus, a proton in the middle of a
-CH 2- chain is likely to have a different frequency from the proton at the
end of the chain. Such a frequency shift, called chemical shift, can be
characterized by a local shielding constant s that modifies the applied
magnetic field as follows:
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
96
Blocal = Bo( 1 - s) Equation 4.1
The value of the shielding constant can be calculated in simple
systems from the electron charge distributions surrounding the nucleus,
but in complex molecules other factors such as electron currents in the
entire molecule have to be taken into account. The shift itself is measured
as the distance in frequency from that of a reference compound. The
chemical shift is thus frequency dependent, as can be seen from the above
expression. To make it invariant with respect to the applied field it is
expressed as a dimensionless number, the chemical shift, given by the
ratio:
8 = 106 [(Fnuc - Fr e f) / Fr e f] Equation 4.2
where Fn uc is the resonance frequency of the sample and Fref is the
resonance frequency of a reference sample. The chemical shift d is
expressed in ppm (parts per million) and is independent of the frequency
of the magnet used. These chemical shifts are rather unique, and by
measuring their values molecules can be identified and the position of a
19
nucleus within a molecule be established. Thus, a F nucleus, say in 5FU,
can be followed in principle as the molecule or its products moving along
metabolic or catabolic pathways in a living system.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
97
4.1.4 SPIN-SPIN COUPLING
In MRI spin-spin coupling is in some sense a nuisance contributing
to the loss of spin coherence and increasing T2 decay. In MRS spin-spin
coupling is a blessing as it permits the probing of the fine structure of
molecules. Spin-spin coupling is the exchange of spin between nuclei
that is not propagated through space but is mediated by the binding
electrons that connect the nuclei. This coupling (usually much smaller
than the chemical shifts) has quantum mechanical implications. Two
identical spin 1/2 nuclei can exist in the following states with respect to
B o: up-up, down-down, up-down, down-up. The last two are degenerate so
that a triplet of states is created each of which can produce a unique
spectrum leading to a triplet peak.
4.1.5 OTHER NUCLEI
The concept of spin-spin coupling applies to all nuclei with spin 1/2
19
in the same manner. For example, two F nuclei can spin-spin couple
together and create first-order splitting, as can two 31P nuclei. In
addition, such nuclei can spin-spin couple with their neighbors to
produce additional fine splitting of peaks in the spectrum. As can be
imagined, if the MRS equipm ent has a high enough resolution, the spectra
from molecules can be extremely busy and complicated. For these cases,
there exist techniques for decoupling spin-spin interactions. When
decoupling is applied to identical nuclei, it is called homonuclear
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
98
decoupling; when applied to different nuclei it is called heteronuclear
decoupling. The method for decoupling a signal is to radiate the coupling
nucleus with a secondary radio-frequency pulse which essentially
prevents that nucleus from spin-coupling with the magnetic absorption
signal of interest. This is known as a spin-coupling suppression.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
99
4.2 5 -FLUOROURACIL (5FU)
The General Introduction, Section 1, discusses in more detail the
properties of fluorine, 5FU and its biochemistry. The following brief
rationale for using F-MRS is based on the nature of the fluorine isotope 1 9 F
1 9
which has favorable NMR characteristics: F is 100% natural abundance,
it has 85% of the sensitivity by comparison with 1H at constant field, in the
mammalian body naturally occurring fluorinated compounds, and it has a
relatively wide chemical shift range, greater than 200 ppm. For a more
general and extensive discussion of 5FU and its byproducts, as well as the
biochemistry, mechanism of action, pharmacokinetic considerations and
distribution in the body following iv injection, please refer to the General
Introduction of this work.
4.2.1 5-FLUOROURACIL
5FU was chosen for our work for several reasons as outlined in the
General Introduction. To summarize:
1. 5FU has been, widely used for about 40 years, and is still viable as a
chemotherapeutic agent in clinical situations. Because of its
widespread and intensive use 5FU the mechanism of its action has been
well characterized and investigated.
19
2. Our work was focused on F MR Spectroscopy, and 5FU contains one
fluorine substituent with a well-defined MR singlet signal.
3. The 5FU metabolites and catabolites have fluorine signals sufficiently
separated from each other, and from the main 5FU signal, yet they do
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
100
not exceed the chemical shift range capable of detection by the MRS
instrum ent.
4. All fluorine detected represented the full dosage administered. Since
the 5FU metabolites and catabolites are also well-characterized there is
little chance of acquiring a spectrum which presents a confusing
picture. Most of the natural fluorine in our body is locked in dentine.
Any fluorine in the body from ingestion of fluorinated drugs or food
which may conceivably add to the fluorine spectrum is generally
verified with a background spectra, that is, prior to 5FU injection.
5. One of the draw-backs of 5FU is that it has a low LD5 0, and has several
side-effects. Also 5FU is not used for types of cancer for which this
particular chemotherapy is contra-indicated.
All the patient data obtained for this work is representative of our
two experimental fields. One field concerns 5FU, and the second field
involves gadolinium diethylene-triamine-pentaacetic acid (Gd-DTPA) used
as surrogate marker, or mimic, of 5FU. Gd-DTPA will be discussed in detail
in Section 5.
4.2.2 TRAPPING
After administration of 5FU the increase of the signal was
monitored in the area of the tumor. The intensity of the signal in the first
spectrum acquired was an average taken over 4.3 minutes. The profile of
the signals during the 4.3 minute averaging period generally consisted of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
101
a rapid rise followed by a gradual exponential decrease. During the
consecutive periods of spectral acquisition the signal intensity tended to
show continued diminishment due to the elimination of the injected 5FU.
From the acquired consecutive spectra we obtained a measure of the
amount of 5FU collected at the tumor site. The retention time of the 5FU
was obtained from a half-life determination, exhibited by the appropriate
signal. From both the half-height, and from the half-area of the initial
peak or area the signal intensity was obtained and both data compared for
deviations. In the case of a noisy background it was more accurate to
utilize the half-area as an index. A half-life in excess of twenty minutes,
as deduced from our experience, usually indicated a so-called “trapper”. A
trapper signified that 5FU elimination form the tumor interstitial fluid
was governed by a low reversal rate. Trapping is beneficial to the patient
because the effective exposure time of the drug to the tumor is essentially
increased. Our experience with over one hundred patients in this
protocol, indicates that roughly 50% of patients are trappers. The
remaining patients were non-trappers. This means that 5FU was within a
retention time range of zero to 20 minutes. By zero we mean below a
detectable limit of the MRI instrument (cf., General Introduction Section
1.4.2).
From the DEMRI experiments (as shown in Section 5) the degree of
vascularization of the tumor site was established. Since the vascular
processes to the tumor are leaky, the drug may enter into the tumor
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
102
interstitial fluid and thus gain access to the tumor cells. The longer the
drug remained in the interstitial fluid the more likely it was that the
tumor cells would take up the drug. The t 1/2 values gave a measure of the
elimination rate of the drug from the tumor interstitial fluid back into the
vascular space. In spite of the fact that a patient is considered a trapper,
this patient may not respond to the 5FU chemotherapy because the tumor
cell may lack the specific enzyme and mechanism required to transport
and metabolize the drug. The latter assumption is presently under study in
our laboratory. Such a patient would be classed as a non-responder.
The same patient was used as a control, i.e., as a baseline prior to 5FU
injection. A variation exists from patient to patient because one may have
a primary tumor different from the other, even in the same organ. The
metastasized tumors may exhibit wide variability, thus leading to a
divergence of backgrounds in different patients. Even in the one liver
the tumor location, size, and age may vary greatly. The sex, age and
condition of the patient are also contributory factors in the variability. A
metastasized tumor located close to a major blood vessel exhibits greater
angiogenesis. In contrast, tumors near tributary blood vessels have less
neo-vascularization (angiogenesis). Thus, the former tumors are much
more aggressive, and the likelihood exists that their vessels are more
leaky.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
103
Because we were concerned with human patients we could not do
more than protocol permits, i.e., blood samples, biopsies, other invasive
techniques for absolute quantification and diagnoses could not be
executed. For this reason, even federal drug administration (FDA)
approved perflurohydrocarbons, utilized as FDA approved blood
substitutes, e.g., Perfluorobrom, may not be used to image vascularization
in the human tumor.
At present access to absolute concentration of the
chemotherapeutic drugs are obtained by means of invasive methods. A
further complication is due to instrument limitations, particularly that
due to inhomogeneity of the area being examined. To wit, the presence of
intervening fat, muscle and skin tend to mask absolute values expected of
the tumor site, i.e., the proton density measured reflects this masking
inhomogeneity. In the case of 5FU fluorine spectroscopy, one attempts to
measure the intensity of the fluorine signal by comparing this signal
with a standard, difluorobenzine (DFB), which is located in the center of
the surface coil. The DFB and 5FU fluorine signals are acquired
simultaneously, and can therefore be read, processed and compared. DFB
as a standard allowed a check on possible frequency shifts during signal
acquisition, DFB also established a peak height, or amplitude, which was
used to normalize the instrument variability in all the components of the
instrum ent.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
104
4.3 METHOD
1. The MR Spectrometry protocol begins with patient qualifications by
an oncologist.
2. If the patient is deemed to be a proper subject for spectral analysis
an iv line is inserted into the vein of the arm.
3. The bolus of 5FU dosage is calculated from the patient’s body surface
area (600 mg 5FU/m2).
4. The subject is placed in the MRI table (Siemens 2.0T Magnetom) and
coronal, sagittal and axial images are acquired to locate the tumor
position and size. The general view of the tumor area is verified
against that obtained from a previous study (CT scan or MRI).
5. The STL coil is now placed on the chosen area, and the coil tuned
and matched for optimum coil performance.
6 . Another image is now obtained for STL surface coil placement.
7. The shim coils in the system (main) magnet which increase the
magnetic field homogeneity are adjusted to yield the highest full
width half maximum (FWHM) value and thus provide the best local
magnetic homogeneity, i.e., to within 50 ppm.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
105
8 . Now background spectra are acquired without 5FU to ascertain that
no background fluorine is present in the patient. These
background spectra are also used for post-data analysis.
9. The 5FU bolus is than injected iv, and the first spectra acquired at
time zero. (MR sequence name is FID-AHP)
10. Subsequent spectra are obtained at intervals of 4.7 min to yield an
average of 10 - 15 spectra, or until the patient wishes to quit the
experiment.
11. The spectral data are analyzed by the algorithm built into the
system computer, and then saved and transferred to a disk drive for
post-processing.
12. Some 120 patients were investigated
13. Post-processing was done with a Power Macintosh computer with
“Mac-FID” as signal processing software, and t 1/2 equations and
graphs derived with “KaleidaGraph” packaged software with the
Wolf (1999) equation.
t 1/2 = mx exp(-m 2 X n^) Equation 4.3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
106
4.4 RESULTS
The spectra obtained were summed 2-dimensionally, i.e., directly
superimposed to provide an average and, thus, for the observation of
hidden signals. See Figure 4.1. The spectra are also presented to the
viewer in a three-dimensional sequential order slightly offset from each
other for the purpose of recognizing patterns. See Figure 4.2. The latter
spectra ipso facto are more “noisy”. Also, patient data involving patient
ID, date of MRS study, sex, age, study type, tumor type and sites, ti/ 2, type of
treatment, and trapper/nontrapper are shown in Table 4.1.
DFB
S-FLi
PATIENT SV132
FBAL
& -u
Summed Patient Spectra of 5FU
Figure 4.1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
107
p ilo t o l th e I spv«.*f r;t .oi‘:p » r ic n t SV 9 1 .
r-t
2D Stacked Patient Spectra of 5FU
Figure 4.2
-R atio A reas FID 5FU SV 101
0.6
0.5
8
td 0.4
I;
■ M
Q
0.3
1
0.2
0.1
y = m !* ex p (-rri? "M0)
Value E rro r
nil 0 .5 9 4 3 1 4 7 8 7 1 4 0 .0 4 1 6 5 6 7
m2 0 .0 2 4 4 1 5 1 9 6 8 6 1 0 .0 0 4 3 5 7 4 2
Chisq 0.01 5 0 5 4 8 5 4 3 8 NA
R 0 .9 3 1 9 3 1 5 2 1 5 4 NA
20 30
Time
Plot of 5FU 2D Stacked Spectra
Figure 4.3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
108
□ Responders
n Non Responders
mmy:
Trappers Non-Trappers
Bar Graph of Data in Table 4.1
Figure 4.4
Table 4.1 Association of Tumor Trapping of 5FU with
Patient Response to Chemotherapy
TRAPPING PATIENT RESPONSE
OF5FU PR NR INEVAL
YES 17 6 1 0
N O 0 29 1 0
5FU bolus (600 mg/m2) with Leucovorin p<0.000001
Trapping: tumoral t1/2 of 5FU is 20 min or more.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
109
4.5 DISCUSSION
The data showed that the patients injected with 5FU could be
classified into two major groups: trappers and non-trappers. The num ber
of patients in this field just exceeded 1 0 0 .
The identification of trappers and non- trappers depends on two
aspects:
1. trapping was determined on the basis of an equation
developed by W. Wolf (1998),
2. duration of 5FU residence in the tumor was taken to be 20
minutes or more (reference private communication W. Wolf
)•
This definition strictly refers to the following categories:
a) The responder and,
b) the non-responder determined by radiologists and
oncologists involved with the chemotherapeutic regimen.
a) T rapper/responder refers to tumor shrinkage obtained by at least two
sets of consecutive images during the chemotherapeutic regimen over
a time span much larger than the time interval of the single data
acquisition experiment. The time span between pre- and post-
therapeutic determination ranges from days to months dictated by the
predeterm ined chem otherapeutic regimen.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
110
b) The trapper/non-responder is characterized by the lack of tumor
shrinkage.
Non-trappers do not respond at all to the 5FU chemotherapy
regimen. The reasons for this may be that the drug does not get to the
tumor site, or if it does access the site its residence time and concentration
are not sufficient to be effective. For trappers the drug is present in the
tumor interstitial space, and most likely has entered the tumor cell proper.
In the case of trappers of the non-responding kind there exist a num ber
of explanatory possibilities, i.e., (1) lack of 5FU-transport into the tumor
cell despite the fact that the drug occupies the interstitium for a
considerable length of time and in sufficient concentration; (2 ) while the
5FU actually entered the cell a mechanism of interaction is lacking. The
latter may involve a deficiency of enzymes, or some other mechanistic
processes as yet undetermined, or not understood.
4.5.1 5FU METABOLITES
The MRS study showed that the 5FU was converted to other
fluorinated derivatives (Figure 4.1), e.g., F-BAL, See example of delayed
spectra. These results could not be correlated with the classes of
responders or non-responders. The appearance of F-BAL and other
metabolites in the delayed spectra confirms the involvement of enzymatic
processes. For patients who did not exhibit the presence of the metabolites
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I l l
this may be due to an actual absence of appropriate enzymes or their
concentration may be below the detection limits of our technique.
The limitation of this spectral method is its relatively poor
sensitivity. The smallest readily workable tumor size was approximately 2
cm in diameter. In terms of the detectability of the tumor size the
diameter was about 0.5 cm. The detectability' of the tumor is limited partly
by the concentration of 5FU. One can actually recognize the presence of
tumors smaller than 2 cm, i.e., down to a diameter of 0.5 cm but these do not
yield sufficiently useful 5FU signals. With other words, the signal is buried
in the background noise, and in order to filter the noise the experiment
has to be extended to an impractical extent of time. Other factors limiting
detectability follow.
The location of the tumor is an important factor.
1. The further the tumor is from the magnetic field iso-center
the more it suffers from deleterious magnetic heterogeneic
factors .
2. We determined that the criterion for optimum data
acquisition was a FWHM to be no greater than 120 PPM.
3. In addition, the tumor needed to be close to the outer
periphery of the patient because proximity of the tumor to
the STL coil allowed less interference for the environment
contiguous to the tumor which introduces contaminating
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
112
signals. The more distant the tumor is from the STL coil the
larger diameter coil must be used. The signal drop-off is
approximately proportional to the square root of the coil
19
diameter. A large coil introduced more F- signal
contamination from non-tumor areas.
4.5.2 5FU LIMITATIONS
Amongst several of the drawbacks present in our 5FU spectrometry
studies in general were, firstly, 5FU although proven to be relatively safe
for patient use at the current dosage, is still to be regarded as a toxic
substance. Secondly, 5FU was limited in its ability to provide clear
information on the extent of vascularization of the tumor. Neither could
one acquire information readily on the leakiness of the vasculature. The
19 19
F signal of the 5FU in the vascular space, and the F signal in tumor
interstitial space cannot be separated or distinguished. Because the time
interval involved to complete a 5FU spectrometry study often exceeds 2
hours, some patients need to discontinue the study because the physical
discomfort of the patient is no longer tolerable. This results in early
termination of the study which in turn leads to lack of sufficient data for
full analysis. Thus, the introduction of a surrogate marker that obviates
these particular 5FU limitations is desirable, i.e., by means of a surrogate
marker which takes the place of 5FU provided that the surrogate markers
are appropriate mimics of 5FU in the patient.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
113
SECTION 5 (DEMRI)
5.1 DYNAMIC ENHANCED MRI (CONTRAST AGENT)
5.1.1 THEORY OF DEMRI
The following information is heavily indebted to a review by Runge
and Nelson. This Section concerns the use of a contrast agent which has
many inherent advantages coupled to non-invasive magnetic resonance
imaging. This can be achieved with gadolinium (Gd) chelate
administration (proton resonance). The benefits of such paramagnetic
contrast agents have increased exponentially, and dictated the method of
choice in the magnetic resonance imaging study presented in this thesis.
Compared with conventional radiography and CT, the mechanisms
responsible for contrast enhancement in MRI are not singular but
multiparametric. The large inherent differences of signal intensity
between various tissues are what make MRI unique compared with other
imaging modalities used in radiology today. In addition, the appropriate
selection of operator-dependent imaging parameters is critical so that
these signal intensity differences can be exploited to optimize MR image
contrast.
The parameters that determine MR signal intensity and contrast are
many (Wehrli, 1991). The first of these, is spin density. Spin density refers
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
114
to the fraction of protons that exists in the voxel of tissue being images
and determines the maximum potential MR signal intensity that can be
realized from that volume of tissue. Most protons that are associated with
organic compounds in tissue. Because the in vivo water content of tissue
cannot be easily altered by a contrast agent, compounds that affect spin
density have received little attention.
Another common param eter exploited in generation of MR contrast
is relaxivity. There are two relaxivity parameters that are unique to each
tissue, Ti andT 2. Longitudinal or spin-lattice relaxation time, known as Ti,
refers to the amount of time it takes for the tissue magnetization to return
to its equilibrium state in the longitudinal direction of the main magnetic
field after excitation with a radio frequency (RF) pulse of energy. The
excess energy that is absorbed by the magnetic spins from the RF pulse is
transferred back to the environment or lattice during the relaxation
process. The second relaxivity property of tissue is the transverse or spin-
spin relaxation, referred to as T2 relaxation. In this relaxation process the
excess energy deposited in the tissue by the RF pulse is transferred
between the magnetic spins. This transferred energy results in loss of
spin phase coherency in the transverse plane and spin dephasing.
Contrast-agent enhancement that is based on alteration of these two
relaxivity parameters can be categorized according to the relative change
it imparts on either Ti or T2 (Engelstad and Wolf, 1988; Gore, 1991; Lauffer,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
115
1990). A contrast agent that predominantly affects Ti relaxation is referred
to as a positive relaxation agent because the enhanced shortening of Ti
relaxation results in increased signal intensity on a Ti-weighted image. By
comparison, a contrast agent that predominantly affects T2 relaxation is
referred to as a negative relaxation agent because reducing T2 results in
decreased signal intensity on a T2-weighted image.
Another determinant of signal intensity in the MR image is
magnetic susceptibility. Susceptibility describes the ability of a substance
to become magnetized in an external magnetic field (Saini et al, 1988).
There are four categories of magnetic susceptibility. Most organic
compounds are diamagnetic substances and have a small, negative
magnetic susceptibility when placed in and external magnetic field.
Paramagnetic and ferromagnetic material have very large net positive
susceptibilities. Diamagnetic susceptibility has a negligible effect in
clinical MRI, and therefore diamagnetic substances are of little interest as
contrast agents.
Paramagnetic substances afford the greatest flexibility in contrast-
agent design and have therefore received the greatest flexibility in
contrast-media development. The presence of a paramagnetic ion can
strongly influence the relaxation properties of nearby protons, leading to
changes in tissue contrast. Paramagnetic contrast agents are
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
116
predominantly used as positive T i relaxation contrast agents, with little
effect seen on T2 relaxation and then only at high concentrations. The
positive net magnetic susceptibility on a paramagnetic ion has little
influence as an actual enhancement mechanism in conventional MRI.
By comparison, the large net magnetic susceptibility of
superparamagnetic and ferromagnetic compounds more directly
influences tissue contrast, with little effect on relaxation per se.
Superparamagnetic substances are individual particles that are large
enough to be a domain. When these particles are exposed to an external
magnetic field, they align with the field, resulting in a large net positive
magnetization. When removed from the magnetic field, they return to
random orientations and lose their net positive magnetization. By
comparison, ferromagnetic compounds are large collections of interacting
domains in a crystalline matrix. They exhibit an extremely large net
positive magnetization in an external magnetic field and maintain this
when removed from the field. Both superparamagnetic and ferromagnetic
compounds have received substantial attention recently in regard to their
application as clinically useful MR contrast agents. These agents function
as negative contrast agents because their large net positive magnetic
moments induce spin dephasing in tissue, with resultant signal loss.
The final two parameters that provide image contrast in MRI are
diffusion and perfusion. The intensity of the MR signal is based on the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
117
magnitude of the bulk magnetization lying in the transverse plane. It is
maximal when all the transverse spins are in phase coherence. Movement
or diffusion of bulk water between tissues in a random motion leads to spin
dephasing and loss of phase coherence in the transverse plane.
Subsequently, this results in the loss of MR signal intensity. Similarly,
perfusion of blood in the microcirculation of the tissue being images also
contributes to spin dephasing and a decrease in the intensity of the MR
signal. In this manner, different degrees of diffusion and perfusion
within tissue contribute to contrast in the MR image. The use of a
relaxivity or susceptibility contrast agent to manipulate diffusion
coefficients (and thus function as a contrast agent) has received limited
attention to date. The presence of a susceptibility agent in the blood pool
can cause large changes in signal intensity. This approach is being
actively investigated as a means of contrast enhancements, specifically
for the measurement of rCBV.
Nearly all the attention on pharmaceutical MR contrast agents has
focused on the use of paramagnetic compounds such as gadolinium
chelates; in our case gadolinium diethylene-triamine-pentaacetic acid (Gd-
DTPA).
A suitable surrogate marker is Magnevist, (brand of gadopentetate
dimeglumine) which is the N-methylglucamine salt of the gadolinium
complex of diethylene triamine pentaacetic acid (Gd-DTPA) manufactured
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
118
by Berlex™ Laboratories, Wayne, New Jersey. This compound is an
excellent contrast agent for MRI experiments. Such contrast agents are
valuable for dynamic enhanced MRI (DEMRI) which allows imaging of the
vasculature of tumors by means of a so-called first-passage phenomenon.
This will be discussed a bit later. DEMRI permits evaluation of vasculature
leakiness into the interstitial fluid space, hence allows a distinction to be
made between signals from the vascular space and tumor interstitial space.
The time required for a complete DEMRI study is approximately 20 min, i.e.,
roughly 10% of that of a 5FU experimental study.
Basically a DEMRI experiment can be divided into two modalities.
One is known as ICAR, initial contrast agent relaxation. This allows one to
determine the extent of tumor vasculature. The other modality, known as
DCAR, delayed contrast agent relaxation, allows the determination of
relative tumor interstitial osmotic pressure.
5.1.2 GADOLINIUM CHELATE
Gadolinium, a rare earth (lanthanide) metal, and therefore not
found naturally in living tissue, has a spin quantum number of 7/2,
making it a desirable relaxivity contrast agent. Gd chelates developed for
MRI have good general safety profile in terms of acute toxicity. Adverse
events encountered are mild and transient, although some very rare
anaphylactoid reactions have been recorded (Carr, 1994). In any chelate
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
119
preparation the addition of a small amount of excess ligand diminishes the
potential for metal-ion substitution (Gennaro et al, 1990).
5.1.3 GADOPENTETATE DIMEGLUMINE (Gd-DTPA)
The chemical structure for DTP A, the chelate in Gd-DTPA (Magnevist®,
Berlex™ Laboratories) is shown in Figure 5.1
The chemical structure of DTPA was the first extracellular
gadolinium chelate to be developed for clinical use. It was approved for use
in the United States by the FDA in 1988 (Goldstein et al, 1990). Currently
this approval includes use in adult and pediatric patients (older than 2
years of age) at a single dose of 0.1 mmol/kg. The pharmaceutical
preparation contains 0.2% of the excess ligand. There is extensive
experience with this agent. As with any such agent, caution should be
exercised in renally impaired patients. The safety of the agent depends to a
large extent on its rapid excretion. Gd-DTPA is cleared by dialysis. In the
coo"
Gd-DTPA
Structure of Gd-DTPA
Figure 5.1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
120
design of an MR contrast agent (Kirsch, 1991; Nelson and Runge, 1989),
such as Gd-DTPA, the first criterion is the ability to alter the parameters
responsible for image contrast in clinical MRI. MRI is unique in that there
are multiple parameters responsible for signal intensity. The contrast
agent is efficient in its ability to influence these parameters at low
concentration to minimize dose and potential toxicity.
This contrast agent possesses tissue specificity in vivo so that it is
delivered to a tissue or organ in a higher concentration than to other
areas in the body. Additionally, once delivered to the desired tissue or
organ, it remains localized for a reasonable period of time so that imaging
can be performed.
The contrast agent is substantially cleared from the targeted tissue
or organ in a reasonable period of time, usually several hours after
imaging, to minimize potential effects from chronic toxicity. The contrast
agent is also excreted from the body, usually by renal or hepatobiliary
routes.
This contrast agent has low toxicity and is stable in vivo while being
administered in doses that can affect the MR relaxation parameters
sufficiently to result in visible contrast enhancement on the MR image.
This contrast agent has little potential for mutagenicity, teratogenicity,
carcinogenicity, and immunogenicity.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
121
In body imaging the high intrinsic tissue contrast of MRI is
notable, lessening to some degree the utility of contrast media.
Administration of a gadolinium chelate is often reserved for specific
problems, such as the differentiation of necrotic from viable tissue, the
identification of active infection, the differentiation of benign from
malignant disease, and the identification of recurrent neoplasia. Specific
applications exist for disease involving the breast (Anzai et al, 1994;
Caramia et al, 1994; Heywang and Kobrunner, 1994; Kerslake et al, 1994)
liver and spleen (Laniado eta/, 1988), kidney (Semelka et al, 1991), pelvis
(Hricak et al, 1995; Hricak and Kim, 1993; Neuerburg et al, 1989) and
musculoskeletal system (Erlemann et al, 1989). Our study allowed the use of
Gd-DTPA to examine tumors of the liver and their micro-circulation by
means of non-invasive MRI techniques.
5.1.4 DEMRI PROFILE
Post-processing yields ICAR and DCAR parameters in terms of voxel
intensity versus time. Voxel refers to a selected region in the acquired
image.
Figure 5.2 is a typical representation of a DEMRI experiment which
for the purpose of this introduction will be divided into 3 portions: Pre
contrast; ICAR; DCAR.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
122
187.39
67.47
0.00
Graph of Patient’s DEMRI 120 pts
Figure 5.2
5.1.4.1 PRE-CONTRAST
The contrast portion (A-B) Figure 5.2 is essentially divided for two
reasons. The first reason is that a steady-state background image without
the contrast agent needs to be acquired. This has to be done because the
lead images (voxel intensity) of the area of interest is brighter than the
subsequent steady state images in the A-B region. The reason for this is
that when the area of interest receives the initial radiofrequency (90
degree) pulse the relaxation signal is at a maximum. Subsequent 90° pulse
signals decrease in intensity compared to the initial 90° pulse signal
because they have less chance to relax fully. After 2 or 3 pulses, the image
signals reach a steady state, the horizontal portion of the pre-contrast
region in Figure 5.2. The second reason concerns post-processing of the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
123
data. Once the steady-state (pre-contrast) has been established it can than
be used for comparative pre- and post-contrast data analysis. At the end of
A but before B , iv bolus injection of the contrast agent is commenced.
5.1.4.2 ICAR
In Figure 5.2 (B-C) the ICAR step represents a first-pass
phenomenon. This means that the contrast agent is present in the
vascular space of the area of interest and has had no chance to migrate out
of this space in the given period. At point B, the start of ICAR, the contrast
agent is first perceived, and in subsequent, images the concentration of
contrast agent increases steadily until a maximum is reached (point
C T max), for the sake of simplifying analysis we assume that in the time B-C,
no contrast agent is lost from the vascular space. Region B to C is an index
of the degree of vascularization. The slope of this part of the curve
indicates the extent of tumor vasculature; the steeper the slope the more
extensive is the vasculature or angiogenesis. At CTmax we believe that the
contrast agent reaches equilibrium in its distribution between the tumor
vasculature and arterial blood concentration.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
124
-
- • © -
- Slope Tum or A
■Slope Tumor B
■ Slope Liver
. Slope Vessel B
r
£
I
200
150
100 -
50
25
y = -70 .2 0 9 + 3.8Q68X R = 0 .9 1 3 1 5
— -y = 5.2927 + 1.2586X R= 0.84608
— -y = -295.03 + 12.118xR= 0.98598
y = -900.59 + 33.122X R= 0.93477
0-0
S l o p e s S V 1 3 2
e— s
-«— I- « -------------- « ■ - ------ » « — » ------1 »
30 35
Image Number
40 45
Graph of Patient’s ICAR Slope
Figure 5.3
5.1.4.3 DCAR
After the maximum, CTmax, the curve decays (region D Figure 5.2)
and DCAR can be evaluated. The rate of decay is an index of tumor osmotic
pressure. Since the tumor has inadequate lymphatic drainage system,
waste products accumulate in the tumor interior and causes a rise of its
osmotic pressure. The more rapid the decay observed in region (D)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
125
signifies a greater tumor osmotic pressure, which indicates that the
contrast agent is eliminated from the tumor interstitial space at a faster
rate. High osmolarity reduces ready access to the tumor by the drug.
■ e— Tumor A
■S - Tumor B
- Liver
C o m p a r a t i v e D a ta S V 1 3 2
O - • V essel B
250
200
150
100
50
20 24 28 32 36 40 44 48
Image Number
Graph of Patient’s DCAR Slope
Figure 5.4
5.1.4.4 MATHEMATICAL MODEL
A solid tumor is spatially heterogeneous with large differences in
the vasculature and in the cells between different regions. The center of
the tumor may contain a necrotic core in which most of the cells are dead.
The outer region of the tumor has rapidly dividing cells, a large blood
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
126
supply, and an abundance of exchange vessels. Another assumption in
these models is that the tumor is spherical. This models also assumes a
spatially homogeneous tum or without lymphatics or extravascular
binding.
The pharmacokinetic model (Wolf Model) employed for analysis of the
acquired dynamic MR image series is shown in Figure5.5.
K
Tum or blood pool
F V ( '
Tumor blood pool
F V
1 > v ll.n* v i
Wolfs Compartmental Model
Figure 5.5
K0 bolus of contrast medium over -30 seconds
kel elimination constant from body blood pool
tdelay
individual lag time from injection
V blood body blood pool volume
Vliver
the intravascular space of liver
F hepatic blood flow
Cl relative conc. of contrast agent in the body blood volume
c 2 relative concentration of contrast agent in the intravascular
space of liver
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
127
Since total acquisition time of the data is approximately 5 minutes
after the administration of contrast medium, the decay (elimination) of Gd-
DTPA in plasma levels can be described by a mono-exponential function.
Therefore, a linear one-compartment open model with zero-order input
(infusion rate, kinput) and first-order elimination (rate constant, kei) was
used to describe the transport of the extracellular contrast medium
through the intravascular space of the liver (liver volume, Viiver)- The
concentration C2 of the contrast agent in the second compartment is
determined by the hepatic blood flow F and by the concentration Ci of the
supply.
Based on this model and assuming a linear relationship between signal
enhancement and mean tissue concentration of the contrast medium, an
explicit expression for the time-dependent signal enhancement observed
in the liver is obtained.
E q u a tio n 5,1
Where kp=F/Vuver is the perfusion rate (in m in 1) and A an amplitude
reflecting the degree of the relative MR signal enhancement. Based on
this equation, the four model parameters, kp, k ei, t^eiay, and A can be
determined by least-squares fitting of the experimental data.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
128
One of the most formidable impediments to the effectiveness of drug
therapy in solid tumors has been the task of overcoming the osmotic
pressure within the tumor in order to deliver the drug. To be successful, a
blood-borne chemical must be distributed through the vasculature, reach
the tumor, cross the vessel wall, and finally diffuse through the
interstitium of the tumor (Jain, 1998). A great deal of research has
therefore been devoted recently to angiogenesis and discovering the
properties that govern the perfusion of blood vessels in tumors. What has
not been so thoroughly examined, however, is the role that osmotic and
high interstitial fluid pressure within tumors affects the rate and volume
of chemical uptake. Physiological studies of tumors have ascertained that
no lymphatic system is incorporated within tumors and thus they have
very poor drainage. In addition, blood vessels generally have a higher
permeability within the tumors, creating a very high and heterogeneous
interstitial pressure(Baxter and Jain, 1990). Future cancer research needs
to factor in the role of interstitial pressure as a limiting force in drug
delivery. Many studies have attempted to analyze tumor physiology by
using models to measure the perfusion rates and correlating these with
tumor growth as well as the ease at which a blood borne therapy can make
its way through the interstitial compartment. We propose that it is
possible to measure the osmotic pressure, vascularization of a solid tumor
in vivo non-invasively with the use of Dynamic Enhanced MRI (DEMRI).
Through three-dimensional analysis of MRI stacks and measurements of
the perfusion and intensity of an enhancing agent, we believe that four
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
129
categories of tumors may be described based on (1) the Initial Contrast
Accumulation Rate (ICAR) value which refers to the rapid uptake observed
soon after contrast agent injection and is a good measure of how well
vascularized the tumor is and (2) the Delayed Contrast Accumulation Rate
(DCAR) which we are using as a measure of the osmotic pressure’s role:
1. High vascularization, low osmotic pressure of the tumor.
The initial ICAR value is high. As time increases, penetration o f the
enhancing agent is close to the tumor center, and the intensity is
high. (High DCAR)
2. High vascularization, high osmotic pressure of the tumor
The initial ICAR value is high. As time increases, penetration o f the
enhancing agent is deep into the tumor center, and the intensity is
low. (Low DCAR)
3. Low vascularization, low osmotic pressure of the tumor
The initial ICAR value is low. As time increases, penetration o f the
enhancing agent is not as far as first category but further then for
second or forth category to the tumor center, and the intensity is
not as high as 1 but, higher than 2 or 4. (High DCAR)
4. Low vascularization, high osmotic pressure of the tumor
The initial ICAR value is low. As time increases, penetration o f the
enhancing agent is furthest from the tumor center, and the
intensity is the lowest. (Low DCAR)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
130
The above boundary conditions (or categories) yields information
regarding the relative osmotic pressure and relative vascularization of the
tumor. From this analysis we hope to show that a mathematical model
governs each of the four tumor categories. This mathematical model can
then be used to predict how well an agent, or chemical can perfuse into
central and periphery regions of the tumor in high and low pressure
systems.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
131
5.2 M ETHOD
The initial steps of the DEMRI procedure is identical to that
described for the 5FU MR Spectroscopy study (Section 1-4), with the
proviso that now the Gd-DTPA replaces 5FU. The concentration for the Gd-
DTPA bolus is consistent with 0.2 mL/kg (0.1 mol/kg). From here on in the
procedure varies. We now use MR imaging not spectrometry, and we
measure a proton relaxation phenomenon. 80-120 “slices” are
programmed into the system’s computer. This represents images of a
selected slice of the subject in a whole body MRI. The duration of the sum
of slice sequences is 4-6 min, and is a repetition of the data acquisition for
that same region. This leads to temporal information of the selected
region. After the fourth datum acquisition, a 20 sec bolus injection of Gd-
DTPA is commenced. These first 4 images normalize the image signal
intensity. This means that the intensity relaxation of the proton signal
attains a fixed value as background. Now we can use the normalized
background image, thus its intensity, for extracting the Gd-DTPA
enhancement in the region of interest. The remaining sequence of
signals which concern the contrast agent is a measure of Gd concentration
in time. This measure is in direct proportion to the Gd concentration. MRI
as well as Magnetic Resonance Spectroscopy (MRS) was performed at the
Los Angeles Oncologic Institute on about 40 different patients, with many
undergoing follow-up imaging. These images were then converted into a
stack format readable by the program NIH Image. NIH Image is used to
measure the signal intensities and contrast changes of the agent used in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
132
the MRI images. 22 different stacks were then analyzed in the following
manner. A line of approximately 10 pixels was chosen so that a section of
the center of the tumor was incorporated as well as section of the
periphery bordering the “normal” tissue. Measurements of the pixel
intensity (representing the mean agent concentration) were taken at
intervals throughout the stack which, in total, correlates to about 5
minutes. The plots of these measurements were then exported to Microsoft
Excel in order to create a dynamic three-dimensional model of the
perfusion of dye into the tumor. See Figures 5.6 and 5.7.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
133
♦
Axial MRI View of Patient Through the Liver with Tumors
Figure 5.6
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
134
SI 1
Image S6
(Time)
40 Density
SV 138 Perfusion
3D view of selected area of the tumor
Figure 5.7
Post-processing yields ICAR and DCAR parameters in terms of voxel
intensity versus time. Voxel refers to a selected region of the acquired
image. Basically, this m ethod involves obtaining a background image of
the tumor by MRI. The Gd-DTPA contrast agent bolus is injected into the
patient’s bloodstream and a series of MRI images acquired of the same
target region. About 120 images are obtained for time lapse information.
The image data of 30 patients were available for analysis, i.e., to determine
the putative tumor DCAR osmolarity index.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
135
5.3 RESULTS
A typical DEMRI curve of voxel intensity versus time and tumor
profile is shown in Figure 5.7. The tumor axis shows to what depth the
contrast agent has penetrated as a function of time.
Figure 5.7 is an example of liver carcinoma in which a section was
taken from the center of a tumor. The tumor was approximately 2.8 cm in
diameter, and the section was approximately 3.0 cm long. The x-axis
represents the section across the tumor, which in the case of the small
tumor above, was 11 pixels, and covered the tumor diameter and bordered
the periphery region. The y-axis represents the relative concentration of
contrast agent, as measured by MR signal intensity, in each pixel. The z-
axis represents a temporal scale, or the 14 out of 80 images chosen to
represent the entire stack. The data can be complicated by the fact that
NIH Image cannot take into account the process of breathing that is
apparent in the images, which can distort the measurements. Images that
correlated the best to one another, while taking into account respiration,
were then chosen by sight. Fewer images in the later half of the stack
were chosen since the measurements began to level off. In addition, the
focus of these charts is to look at the initial uptake of the contrast agent in
the center of the tumor, as opposed to the periphery and well-perfused
regions. The MRS clinical trials involving 5-flurouracil (5FU) therapy
were performed in conjunction with this study. In these studies the
drug’s half-life, or the time at which half the maximum concentration of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
136
5FU is measured within the tumor area, is examined via spectroscopy.
Relative comparisons of the half-life values can then be used to designate
whether the tumor is a drug “trapper” or “non-trapper” The data derived
in this way serves as the quantitative, in-vivo measurements for which we
can compare predictions formulated by means of a theoretical model.
This 3-dimensional relationship between the depth into the tumor
from the periphery of the tumor and the image intensity as a function of
time should demonstrate osmotic pressure or rather, the difference of the
ionic gradients between the tumor compartment and the vasculature
compartment due to a lymphatic insufficiency which would have
eliminated the debris from the tumor. The apex of each set of peaks
increases with time until a maximum has been reached and begins to
diminish after a given time. The rate of penetration of Gd-DTPA towards
the center of the tumor for a particular concentration signifies the
relative osmotic pressure of the tumor interstitial space. A sampling of
such 3-dimensional plots shows the different characteristics of Gd-DTPA
penetration but only give relative osmotic pressures. Analysis of sets
provided only two possibly useful data sets which could be construed as an
index of osmotic pressure. The reason for this is that in these particular
sets the trailing segm ent o f the DEMRI curve actually had a discernible
negative slope. The result from this set yielded a predicted ti / 2 for a small
molecular weight drug, in this case the Gd-DTPA to be 19.1 min. This
prediction was borne out by the observed ti / 2 value of 20 min for 5FU.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
137
5.4 DISCUSSION
Qualitatively speaking, the figures demonstrated that all the tumors
studied display many degrees of variation of how interstitial fluid pressure
affects perfusion in the center of the tumor as well as the periphery. In
terms of measuring agent uptake, many tumors display a sharp slope at the
periphery, yet a more diminished slope in the center of the tumors. This
would indicate that the agent could not diffuse well into the tumor center.
In other cases, conversely, the tumor center, while having a lesser overall
mean density, shows an almost similar uptake slope for the periphery of
the tumor. This indicates a more thorough diffusion. Another important
factor that must be noted is the size of the tumor(s). One observation that
was made was that in the cases of small tumors, there was nearly uniform
and high uptake, whereas larger tumors displayed more heterogeneous
perfusion, as well as a clear necrotic core. Smaller tumors may not have
yet developed the high interstitial pressure that is seen in larger tumors,
and thus there is greater perfusion. In essence, the absolute size of these
small tumors may be able to offset the absence of a lymphatic system.
Unfortunately the most difficult part of this study has been the
formulation of accurate mathematical models to describe the functions of
interstitial pressure in this heterogeneous perfusion. One o f the main
complications has been finding the exact relationship of perfusion and
pressure with the four categories listed above. We have not yet concluded
whether a two model system can be used to represent the high and low
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
138
pressure tumor classifications, or if they are all defined by slightly varied
models.
Currently, we have implemented the graphs in two cases only
qualitatively by comparing the slopes of the graph to the MRS clinical data
of uptake and half-life of the 5-flurouracil. We found that in this one case,
in which the uptake slope within the center and at the periphery of the
tumor was high, the clinical data for the patients indicated drug trapping
and a relatively high half-life. This indicates a good, albeit qualitative,
correlation between the MRI data and the clinical MRS tests.
Our DEMRI study demonstrated the possibility of gauging the
relative osmolarity of tumors in vivo and non-invasively. This confirmed
work by other investigators that the hepatic tumor lacked an adequate
lymphatic system. Our DEMRI data also showed that Gd-DTPA can not enter
the tum or cell and interact with the endogenous enzymes because there
are no Gd-DTPA transport mechanisms available in the tumor cell
membrane. Although our DEMRI ICAR study excluded enzyme information
it did yield rapidly, in high temporal resolution, tumor vascularization and
osmolarity. In other words DEMRI quantitatively allowed the
determination of residence time in the tumor interstices as an index of low
molecular weight drug presence. The actual osmotic pressure in pressure
units cannot at this stage be determined per se. All we can do at this
juncture is to substitute for pressure a time index. With other words, the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
139
residence of Gd-DTPA can be measured in time as a ti / 2 and then compared
with the 5FU ti / 2 data determined on the same patient. The small data sets
here presented can only give a guideline to the possibility of using DCAR
as a substitute for 5FU spectroscopic studies, but with the benefit of
shortening the experimental exposure. These data sets demonstrate that
Gd-DTPA is a mimic of 5FU with the limitation that Gd-DTPA cannot enter
the tumor cell thus, further work is evident.
6. COMBINED DISCUSSION
The majority of patients subjected to our STL coils were able to
complete the study because the coils were considered to be more
comfortable by patient comment. This comfort index also improved the
STL acquired spectra compared to previous studies because patient
movement was reduced. Further, the shimming time for adjusting fields to
homogeneity was reduced by about 30%, i.e., from 30 min to less than 20
min. Again diminishing patient total study time. With our STL coils we
noted an improvements of S/N ratio by, an average, about 8-10%. With our
STL coils we were able to distinguish spectral features not previously noted
in the older data. The use of these coils facilitated the information
presented below.
The experimental DEMRI data obtained in this work shows that
aggressive tumors have a characteristic angiogenesis, or
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
140
neovascularization. This is demonstrated by the fact that the signal
enhancing molecules like Gd-DTPA concentrated in the corona (outer)
region of the tumor. Gd-DTPA is designed to show micro-vascularization,
and by virtue of its highly charged nature cannot penetrate a cell
membrane. Some Gd-DTPA may leak into the interstitial fluid from the
blood vessel but does not enter the cellular compartment. In a normal
blood vessel, during the first bolus passage no measurable amount of Gd-
DTPA was found outside the blood vessel. As the number of Gd-DTPA passes
increased leakage was detected to a small extent. This implied that even in
the first passage it is likely that some Gd-DTPA will actually adsorb on to
the blood vessel membrane and than be transferred across. Interstitial
fluid detectable Gd-DTPA concentrations were found after, or within, 2 to 3
min. In the human body almost every cell has an interstitial fluid
component, however, liver cells do not. A tumor in the liver is not itself a
liver cell and it contains interstitial fluid. The reason for this is that a
liver tumor cell is generally metastasized from a region other than the
liver. The normal blood vessel in the liver delivers the Gd-DTPA to the
blood vessels of the tumor. The reason the tumor blood vessels leak, as
mentioned previously, is because the cells of the tumor blood vessel grow
at such a rapid pace that the organization of these cells does not develop
normally but tend to assemble very poorly, lacking a proper sealing
structure. In a number of patients DEMRI readily demonstrated this. For
example, in patients SV138 and SV 162. For the patients which do not show
this Gd-DTPA enhancement, we believe that this phenomenon arises
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
141
probably for two major reasons. One is that the tumor is not of an
aggressive nature implying a small amount of neovascularization or
angiogenesis, and the other reason is that angiogenesis is taking place at
such a rate that the tumor blood vessel has no chance to organize into cells
more along the lines of a normal, or regular, blood vessel.
Some investigators (Dr. Degani’s work in their laboratory) used this
phenomenon to distinguish between malignant and benign tumors. This
left open the question whether these characteristics could be used to
predict whether a benign tumor may metamorphose to a malignant tumor.
In the case of smaller lesions discrimination would be much more
difficult. However, a recent study by Degani and coworkers have
demonstrated that DEMRI can be used even in these cases.
Our work concerned mostly the initial stage of DEMRI to avoid
complications due to Gd-DTPA leakage into the interstitial fluid, whereas
Degani’s research was focused on different leakage rates (as properties of
the tum or vessel).
The tumor was recognized by us to consist of three separate regions:
the active or outer regions; the inner, or very liquid, cystic region; and
an intermediate or necrotic, or swampy region, cf., General introduction
(Section 1). DEMRI was capable of showing characteristics of vascular
activity which takes place in the outer region. This is the region where
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
142
vascular leakage may be observed. The swampy region was demonstrated
by us with experimental animals (rats) and by pathologic examination by
other investigators to be in the process of cellular break-down due to
anoxia. With other words, the cells in this region were being choked to
death induced by the lack of nutrients and oxygen. Also, the waste
products are not eliminated or excreted, or do so with difficulty. This is
due to the fact that new cells form a coating on the older, underlying cells,
depriving the older cells from access to required nutrients. Some of the
cells in the swampy region may still be viable. In the inner region,
demonstrated again by pathologic examination, most of the cells were dead
or lysed. DEMRI showed that the outer region of the tumor, when viewed
in section, will initially brighten yielding a corona-like effect. This is
predominantly a vascular effect. The swampy region also showed a
brightening effect but at a much later stage, several minutes later , and
we determined that this was due to a diffusion process occasioned by the
leakage of Gd-DTPA from the outer, or active region. The inner region
exhibited Gd-DTPA brightening very rarely, or never. However, if the
DEMRI experiment was extended to longer than 6 minutes, it is conceivable
that some minor diffusion to this region could occur. The observation of
three distinguishable regions in a solid tumor received attention via
radio-labeling done by a previous worker in our research group . Our
work on tumor sections staining actually demonstrated these regions
unequivocally.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
143
Gd-DTPA was chosen for patient use because it is the only marker
approved by the Food and Drug Administration. Other similar markers can
be used in animal studies, and in vitro work, to simulate or mimic
chemotherapeutic agents. However, our experiments involved chemical
work with human patients. It devolved upon us to demonstrate that Gd-
DTPA does indeed mimicked chemotherapeutic agents like 5-fluorouracil.
The tumor osmolarity experiment demonstrated the same t1 /2 for Gd-DTPA
and 5FU, and thus Gd-DTPA served as a mimic, but not in all respects, i.e.,
with enzymatic and tum or membrane transport.
In order to be a suitable surrogate, a marker must have a set of
characteristics as explained previously. In this context the surrogate
marker did not leave the vascular space readily; it provided a signal easily
detected by the instrumental method, and that in minimal concentration.
DEMRI in the MRI range of spectroscopic methods was found to be the
optimal tool for detecting Gd-DTPA levels, i.e., minimum marker for
maximum effect. Further, the marker did not catabolize, or metabolize or
change or react in any way. Marker toxicity was extremely low, well
below the concomitant lethal dose of the chemotherapeutic drug 5FU. The
marker also did not interfere, with the anti-cancer drug either by
synergism, activation or by inhibition. The surrogate marker was
retained by the experimental subject in approximately the same time
frame as the chemotherapeutic drug.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
144
In our study we needed Gd only to show vascularization. The
requisites of toxicity, dosage, time frame of action, synergy or inhibition
fell within acceptable limits. Tumor vascularization is important because
one needed to know whether the “plumbing” is available for the
chemotherapeutic drug to access the tumor. If the drug is injected
systemically, a vascular route is essential. In the case of skin
melanoma(as an aside), the drug is injected directly to the tumor, hence no
vascularization per se is necessary. In a pilot study we conducted of
malignant skin melanoma we demonstrated that the mode of drug delivery
did not require vascularization. However, in the present study of tumors
metastasized to the liver in human subjects direct injection into the tum or
area was not a practical method. Thus, delivery of the chemotherapeutic
drug to the tumor sites in question required vascularization.
Gd-DTPA is more routinely used to confirm the presence of
suspected tumors in the liver and other areas. Indeed, in the case of brain
tumors this marker is the only, and at present, the best way of making
such a diagnosis. The reason for the effectiveness of the Gd-DTPA
technique is primarily, and probably solely based on the fact that the
neovascularization of the tumor vessels is very leaky. The Gd-DTPA, in
spite o f its highly charged nature, is capable o f passing through the blood
brain barrier and therefore manages to access solid tumors. There the
tumor actually breaks down the blood brain barrier because of its
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
145
parasitic nature. This parasitic process occurs in most solid tumors,
including tumors in the liver.
The MRI image of the tumor was distinguishable from its
surrounding environment by as is commonly known as “lighting up”, or
with other words a change of intensity brightness occurred, with signal
enhancement observed in the suspected pathological mass, while the
contiguous area remained relatively unchanged. The degree of relative
signal intensification was proportional to the concentration of Gd-DTPA
interacting with the suspected tumor site. Gd-DTPA presence also caused a
“brightening”. In normal blood vessels Gd-DTPA was confined in the
initial stage, but a tumor’s blood vessel leakage caused much less
prolonged signal enhancement.
MR angiography (MRA) was conducted by presaturating the MR
(proton) signal in the region of interest. The relaxation of the proton was
deliberately avoided. The second consecutive acquired image of the same
section only the area which had a high signal showed the unsaturated
blood entering the scan region. MRA was unidirectional in terms of blood
flow i.e., it allows distinction between arteries and veins without the use of
contrast agent. This technique did not allowed us to discriminate between
overall density of vascular distribution. However, DEMRI did allow such
an determination by comparing intensities of pre and post image data,
and the change in signal intensity was linearly proportional to Gd
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
146
concentration, i.e., a measure of the distribution density of blood vessels.
Since MRA utilizes no contrast agent it is not possible to determine
leakage, unlike in the case of DEMRI which utilizes an exobiological
marker. For a human patient the resolution capability of the MR
instrum ent allowed only recognition of comparatively large vessels.
Neovascularization was generally of a microscopic size. MRA was
incapable of imaging micro-vessels. Although DEMRI also does not image
individual microvessels, it can give an image of a number of microvessels
summed. With other words, the presence of microvessels can be observed
in extremely small areas of its interest. In addition, the scope of the image
area could be expanded to take in a whole, or large, tumor.
6.1 5FU FIELD
The data showed that the patients injected with 5FU could be
classified into two major groups: trappers and non-trappers. The number
of patients in this field just exceeded 100 (See graph 1). The identification
of trappers and non- trappers depended on two aspects: (1) duration of 5FU
residence in the tumor taken to be 20 minutes or more (reference private
communication W. Wolf ) for trappers, (2) trapping was determined on the
basis of an equation developed by W. Wolf (Equation 4.3). This definition
strictly refers to two categories: (a) The responder and (b) the non
responder determined by radiologists and oncologists involved with the
chem otherapeutic regimen.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
147
a) Trapper/responder referred to tumor shrinkage obtained by at least two
sets of consecutive images during the chemotherapeutic regimen over a
time span much larger than the time interval of the data acquisition
experiment. The time span between pre and post-therapeutic
determination ranged from days to months.
b) The trapper/non-responder was characterized by the lack of tumor
shrinkage.
Non-trappers did not respond at all to the 5FU chemotherapy
regimen. The reasons for this may be that the drug did not get to the tum or
site, or if it did access the site its residence time and concentration are not
sufficient to be effective. For trappers the drug was present in the tum or
interstitial space, and most likely had entered the tumor cell proper. In the
case of trappers of the non-responding kind there exist a number of
possibilities, i.e., lack of 5FU-transport into the tumor cell despite the fact
that the drug occupied the interstitium for a considerable length of time
and in sufficient concentration; while the 5FU actually entered the cell it
lacked the mechanism of interaction. The latter may have involved a
deficiency of enzymes, or some other mechanistic processes as yet
undetermined.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
148
6.1.1 5FU METABOLITES
The MRS study showed that the 5FU was converted to other
fluorinated derivatives (Figure 4.1), e.g., F-BAL, See example of delayed
spectra (Figure 4.2). These results could not be correlated to the classes of
responders or non-responders. The appearance of F-BAL and other
metabolites in the delayed spectra confirmed the involvement of
enzymatic processes. For patients who did not exhibit the presence of the
metabolites this may be due to an actual absence of appropriate enzymes or
their concentration have been below the detection limits of our technique.
The limitation of this spectral method was its poor sensitivity. The smallest
readily workable tumor size was approximately 2 cm in diameter. In terms
of the detectability of the tumor size the diameter was about 0.5 cm. The
detectability of the tumor was limited partly by the concentration of 5FU.
One can actually recognize the presence of tumors smaller than 2 cm, i.e.,
down to a diameter of 0.5 cm but these do not yield sufficiently useful 5FU
signals. With other words, the signal was buried in the background noise,
and in order to filter the noise the experiment had to be extended to an
impractical extent of time. Other factors limiting detectability follow.
The location of the tumor was an important factor. The further the
tumor was from the magnetic field iso-center the more it suffered from
deleterious magnetic heterogeneic factors . We determined that the
criterion for optimum data acquisition was a full width/half maximum
(FWHM) to be no greater than 120 PPM. In addition, the tumor needed to be
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
149
close to the outer periphery of the patient because proximity of the tumor
to so-called the in-house coil allowed less interference for the
environment contiguous to the tumor which introduced contaminating
signals. The more distant the tumor was from the in-house coil the larger
diameter coil had to be used. The signal drop-off was approximately
proportional to the square root of the coil diameter. A large coil introduced
more F- signal contamination from non-tumor areas.
6.2 DEMRI FIELD (Gd-DTPA)
Two features (ICAR and DCAR) of contrast agent DEMRI curves were
useful for predictive distribution of small molecular weight
chemotherapeutic drugs in solid tumors.
6.2.1 ICAR
ICAR of the surrogate marker Gd-DTPA gave an alternate
representation of how suitable drug molecules would be delivered to the
target in a considerably reduced time span compared to other methods. It
also gave an index for tumor vascularization which may not be obtained
readily from a 5FU study. This feature allowed prediction partly why
trappers may not be responders. The importance of vascularization
morphology determined to a large extent the availability of the
chemotherapeutic drug to the tumor. If the tumor is poorly vascularized
the drug concentration and exposure parameters were just be insufficient
even though other factors may be favorable.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
150
6.2.2 DCAR
DCAR may be suitable for the determination of tumor osmolarity
which is important for the entry of the drug molecule into the tumor
interstitial space. High osmolarity reduces ready access to the tumor by
the drug in spite of high tumor vascularization. In addition, high
osmolarity may also reduce drug residence time in the tumor interstitial
space and thus provides a barrier for entry into the tumor cells. For the
DEMRI DCAR study some 30 patient data sets were potentially available.
Two sets of data could be used.
6.2.3 INTERSECTING FIELDS: 5FU/DEMRI (ICAR)
The ICAR data could not substitute for the 5FU data with a high
degree of accuracy, i.e., tumor vascularization was not readily visualized
because of the availability of only one or two data points in the region (B)
of the plot in Fig 5.2. For this reason a higher DEMRI temporal resolution
provided a more satisfactory method for determining tumor
vascularization. The permeability of the tumor vascular wall was shown to
be much greater for 5FU than for Gd-DTPA. Therefore in the first 5FU
spectra the intensity of the F signal was due to 5FU in both the tumor
vascular and interstitial space. The 5FU field represented >100 studies and
was given in terms of responder (R) and non-responder (N) based on 5FU
t1 /2 data. The ICAR values of the DEMRI field represent 30 studies and the
values were given in retention time. We wished to ascertain what
correlation exists between these 2 fields. Gd-DTPA remained in the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
151
circulation volume for approximately 25 min. This value was chosen to
represent the retention time of the marker by the tumor. Any lower time
value may be considered as non-retentive by the tumor. Table 4.2 shows a
column labeled 5FU response indicated by R as responder and N as non
responder. The second column gives retention times in min. The third
column shows whether there was a correspondence between columns 1
and 2. It is noted that for the 18 data sets available from both fields, there
were 13 correlated cases and 5 did not correlate. Thus, a 71% correlation
was found. This tells one that a high degree of vascularization is likely to
lead to positive chemotherapeutic response.
6.2.4 INTERSECTING FIELDS: 5FU/DEMRI (DCAR)
By analyzing the DCAR data one was be able to extract t\/2
information conventionally acquired from the 5FU experiment. The decay
of the DCAR profile may be subjected to the same mathematical model for
the determination of 5FU t \ /2- This is due to the fact that in the tumor
vasculature and interstitial fluid space the Gd-DTPA mimics the behavior
of 5FU, thus serving as a surrogate drug. The possibility of determining
the ti / 2 values via DCAR by direct experiment obviated the reliance on
theoretical mathematical predictions. Such an experimental approach may
well allow discrimination of various mathematical models in terms of their
efficacy. This finding provides leads for further work. If ti /2 values can be
determined easily from DCAR, the time for data acquisition via DEMRI may
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
152
well be shortened significantly, with obvious benefit to the patient. And
the use of a toxic drug for study, e.g., 5FU, is avoided.
A severe limitation of the determination of ti /2 from a DCAR
experiment is that it cannot provide for an appropriate selection of a
chemotherapeutic drug. Gd-DTPA, although it mimics 5FU cannot cross the
tumor cell membrane, thus the tumor enzyme activity cannot be evaluated
by this means.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
153
7. CONCLUSIONS
1. A series of surface trans-locatabie (STL) coils ranging in diameter
from 3 cm to 15 cm were designed and manufactured in-house. These could
be placed on to the patient’s body for maximum comfort and derive better
data acquisition than commercial units. The latter required fitting the
patient to the coil. The signal-to-noise ratio of the STL coil was
significantly increased, and allowed a vast improvement in the magnetic
resonance spectroscopy data acquisition in terms of sensitivity and
resolution. The STL coils were adaptable to most commercially available
versions of magnetic resonance imaging and spectroscopy thus providing
virtually universal applicability. These coils can now be used for multi
center studies presently in progress. Further work on coil improvement
consists of adjustable diameter size which allows the one unit to be
instantly adaptable for different tumor dimensions and location. Such a
coil can be relocated on the patient to different areas without having to
change electronic connection which will also speed up the total
experimental contact time.
19
2. The STL coils were used for F spectroscopic studies successfully in
several medical centers. Over 100 patients were studied, and within the
probability limit of p = 0.0001 we determined that non-trappers were non
responders to 5FU chemotherapy. Thus, providing the oncologists with
data which allowed them to decide that in these cases the 5FU
chemotherapeutic regimen would be ineffective. This could be achieved
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
154
with a single dose of 5FU within a few hours. This technique is
immeasurably more effective than that of the method which the
oncologist had to rely on to determine tumor progression. This also
reduced patient suffering to a minimum, saved precious therapeutic time,
and allowed a rapid early shift to a different regimen. This STL coil
spectroscopic method determined that 48% of the patients examined had
tumors which were trappers of 5FU. However, amongst these trappers 41%
were be classed by the oncologists as non-responders. Hence, for the
remaining 59%, 5FU therapy was indicated (see Table 4.2). To understand
the exact situation with trapper- non-responders further work should
include possible enzyme and drug-transport studies possibly involving
tumor vascularization. With other words no matter how good the
anticancer drug may be, the tumor has to be exposed effectively to that
drug for it to be effective.
3. We developed a non-invasive, in vivo, dynamic enhanced magnetic
resonance imaging technique which allowed the rapid quantification of
tumor vascularization and angiogenesis. By acquiring 3-dimensinal
images (the time domain is the 3rd dimension) we were able to observe the
depth of penetration of a contrast agent, Gd-DTPA, and the extent of its
retention in the tumor. With this information the relative osmotic
pressure of a tumor could be determined. However, the number of data
was small, and in order to use this technique effectively the time of the
experimental study needs to be extended. Further work in this respect is
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
155
required. We established that the 5FU ti /2 times correlated with the Gd-
DTPA ti /2 times. The Gd-DTPA may therefore be used as a surrogate marker
for 5FU or small molecular weight molecules. However, Gd-DTPA does not
penetrate the tumor cell wall and thus, although it is a mimic, it is
incapable of use for enzyme or drug transport study.
4. Current evaluation methods of chemotherapeutic effectiveness is
done by multiple infusion of the drug over 3 to 6 months, and comparing
periodically the size of the tumor by MRI and CT scans. This determines
whether the current chemotherapeutic regime and dosage was useful.
19
With our F-MRS procedure we could determine within a few hours
whether the 5FU chemotherapeutic process will not be effective (non
trapper). If by our methods we determined the patient to be a trapper
further evaluation was initiated. By a rapid evaluation the patient’s
immune system is not as severely compromised. Early evaluation allows a
shift toward a more effective drug regimen.
5 Non-invasive methods of measuring aspects of tumor physiology,
such as perfusion and pressure gradients, are key tools in the
future of cancer therapy. Magnetic resonance spectroscopy clearly
offers beneficial means for clinical and pre-clinical cancer
applications. It is ideal for measuring the availability and the
kinetics of drug delivery to target sites. Using DEMRI analysis in
association with the MRS data provides a very useful tool through
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
156
which we will learn about tumor biology, pharmacokinetic, and
possible prescriptions for therapy. Although the observations and
the results we have gleaned from the MRI and MRS data are not yet
conclusive; one can see the possible clinical applications on the
horizon. For example, as Baxter and Jain noted, monoclonal
antibodies, while being extremely advantageous in regards to being
able to target cancer tissue specifically, have had limited success
because they cannot be distributed evenly throughout the tumor
(Baxter and Jain, 1989). This can be due to the non-uniform blood
vessel perfusion within tumors, or quite possible the role of the
interstitial pressure that we are examining. The same holds true for
the variation in results seen in the data collected from the MR
Spectroscopy and 5FU studies that have been done here. Studying in
what creates this pressure heterogeneity, as well as methods of
alleviating or working with it, may lead to clinical breakthroughs
in overcoming the physiological boundaries that limit many cancer
therapies.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
157
8. FUTURE WORK
1. COIL DESIGN
The STL fixed diameter loop coils have been shown to be superior to
commercially available coils for the reasons cited above. However, it is
possible to construct a single unit for which the coil diameter may be
adjusted continuously to any dimension desired. The principle of such a
variable diameter coil is based on a sliding segmental mechanism which
would allow adjustment of the field size appropriate for tumor size and
tumor location. The design for a prototype is now available but its
construction awaits future work.
Another improvement to the STL coil currently in use can be the
incorporation of a PIN diode in the electronic circuit to tune and match
the STL coil electronically rather than adjusting the capacitance of the
coil manually.
2. TUMOR INTERSTITIAL OSMOTIC PRESSURE
The tumor interstitial osmotic pressure may be influenced by a
hypotensive drug to increase the tum or vascular space thereby increasing
the gradient between the tumors interstitial fluid and tumor vascular
spaces allowing more debris to leave the tumor interstitial fluid space.
Once the altered gradient is established, a chemotherapeutic agent may be
administered with a hypertensive drug simultaneously (to reverse the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
158
pressure gradient), and force greater chemotherapeutic agent perfusion.
The feasibility of this approach may be tested with a corneal implanted
tumor.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
159
9.0 REFERENCES:
Abe I, Suzuki M, Hori K, Saito S and Sato H, Some aspects of size-dependent
differential drug response in primary and metastatic tumors, Cancer
Metastasis Rev 4:27-39,1985
Ahmad I, Longenecker M, Samuel J and Allen TM, Antibody-targeted
delivery of doxorubicin entrapped in sterically stabilized liposomes can
eradicate lung cancer in mice, Cancer Res 53:1484-1488, 1993
Ahmad I and Allen TM, Antibody-mediated specific binding and
cytotoxicity of liposome-entrapped doxorubicin to lung cancer cells in
vitro, Cancer Res 52:4817-4820,1992
Ames MM, Powis G and Kovach JS Eds, Pharmacokinetics of Anticancer
Agents in Humans, Elsevier, Amsterdam, 1983
Anzai Y , Blackwell KE, Hirschowitz SL, et al, Initial clinical experience
with dextrancoated superparamagnetic iron oxide for detection of lymph
node metastases in patients with head and neck cancer, Radiology 192:704,
1994
Armstrong RD, Lewis M, Stern SG, Acute effects of 5-fluorouracil on
cytoplasmic and nuclear dihydrofolate reductase messenger RNA
metabolism, J Biol Chem 261:7366-7371,1986
Baguley BC and Finlay GJ, Pharmacokinetic/cytokinetic principles in the
chemotherapy of solid tumors, Clin Exp Pharmacol Physiol 22:825-828, 1995
Baish JW, Netti P and Jain RK, Transmural coupling of fluid flow in
microcirculatory metwork and interstitium in tumors, Microvasc Res
53:128-141,1997
Balis FM, Holcenberg JS and Bleyer WA, Clinical pharmacokinetics of
commonly used anticancer drugs, Clin Pharmacokinet 8:202-232, 1983
Bapat AR, Zarov C, Danenberg PV, Human keukemic cells resistant to 5-
fluoro- 2 ’ -deoxyuridine contain a thymidylate synthetase with lower
affinity for nucleotides, J Biol Chem 258:4130-4136, 1983
Baxter LT , Yuan F and Jain RK, Pharmacokinetic analysis of the
perivascular distribution of bifunctional antibodies and haptens:
comparison with experimental data, Cancer Res 52:5838-5844, 1992
Baxter LT, Zhu H, Mackensen DG, Butler WF and Jain RK, Biodistribution of
monoclonal antibodies: scale-up from mouse to human using a
physiologically based pharmacokinetic model, Cancer Res 55:4611-4622,
1995
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
160
Baxter LT and Jain RK, Transport of fluid and macromolecules in tumors I.
Role of interstitial pressure and convection, Microvascular Research, v.37,
p77-104,1989
Baxter LT and Jain RK, Transport of fluid and macromolecules in tumors II.
Role of heterogeneous perfusion and lymphatics, Microvascular Research,
v.40, p246-263,1990
Baxter LT and Jain RK, Transport of fluid and macromolecules in tumors
III. Role of binding and metabolism, Microvasc Research 41:4-23, 1991
Baxter LT and Jain RK, Pharmacokinetic analysis of the microscopic
distribution of enzyme-conjugated antibodies and produrgs: comparison
with experimental data, Br J Cancer 73:447-456, 1996
Beatty BG, Beatty JD, Williams LE, Paxton RJ, et al.„ Effect of specific
antibody pre treatm ent on liver uptake of lllln-labeled
anticarcinoembryonic antigen monoclonal antibody in nude mice bearing
human colon cancer xenografts, Cancer Res 49:1587-1594, 1989
Bendall MR, Pegg DT, Effect of thyumidylate synthase in MRI,J Magn
Reson, 67:376-381,1986
Benz C, Cadman E, Modulation of 5-fluorouracil metabolism and cytotoxicity
by antimetabolite pre treatm ent in human colorectal adenocarcinoma HCT-
8 cells, Can Res, 41:994-999,1981
Berger SH, Jenh CH, Johnson LF, et al., Thyumidylate synthase
overproduction and gene amplification in fluorodeoxyuridine-resistant
human cells, Mol Pharmacol 28:461-467,1985
Bhujwalla ZM, Shungu DC, Glickson JD, and van Zijl, In vivo detection of
1 3 C-glucose uptake and 1 3 C-lactate efflux in a murine tumor by 1 H(l3C) NMR
spectroscopy following intravenous infusion of (l-1 3 C)-glucose, 11*
Annual meeting of the Society of Magn Reson, Berlin, p 3536, 1992
Blumenthal RD, Fand I, Sharkey RM, Boerman OC, Kashi R and Goldenberg
DM, The effect of antibody protein dose in the uniformity of tumor
distribution of radioantibodies: an autoradiographic study, Cancer
Immunol Immunother 33:351-358, 1991
Bottomley PA, Charles HC, Roemer P, Flamig D, Engeseth H, Edelstein WA,
Mueller OM, Novel drug delivery, Magn Reson Med 7:319, 1988
Boublil JL,Milano G, Khater R, et al., Continuous 5-day regional
chemotherapy by 5-fluorouracil in colon carcinoma, Pharmacokinetic
evaluation, Dr J Cancer 52:15-20, 1985
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
161
Caramia F, Huang Z, Hamberg LM, et al, Identification of peripheral areas
of cerebral flow-volume mismatch in focal transient ischemia using
susceptibility contrast dynamic MRI, Proc Soc Magn Reson 37:196, 1994
Carr JJ, Magnetic resonance contrast agents for neuroimaging: safety
issues, Neuroimaging Clin N Am 4:43, 1994
Chen C-N, Hoult DI, Sank VJ, Transport of fluid in solid tumors, J Magn
Reson 54:324,1983
Chen HSG and Gross JF, Estimation of tissue-to-plasma partition coefficients
used in physiological pharmacokinetic models, J Pharmacokinet Biopharm
7-1 1 7-1 9 ^ 1Q 7Q
I IXX I
Cheng YC, Nakayama K, Effects of 5-fluoro-2’-deoxyuridine on DNA
metabolism in HeLa cells, Mol Pharmacol 23:171-174, 1983
Cheung NK, Neely JE, Landmeier B, Nelson D and Miraldi F, Targeting of
ganglioside GD2 monoclonal antibody to neuroblastoma, J Nucl Med
28:1577-1583,1997
Clark JL, Berger SH, Mittelman A, et al., Thymidylate synthase gene
amplification in a colon tum or resistant to fluoropyrimidine
chemotherapy, Cancer Treat Rep 71:261-265, 1987
Cohen A, Ullman B, Role of intracellular TTP levels in fluorodeoxyuridine
toxicity, Biochem Pharmacol 33:3298-3301,1985
Collins JM, Pharmacokinetics of 5-fluorouracil infusions in the rat:
Comparison with man and other species, Cancer Chemother Pharmacol
14:104-111,1985
Collins JM, Dedrick RL, King FG, et al., Non-linear pharmacokinetic models
for 5-fluorouracil in man, Clin Pharmacol Ther 28:235-246, 1980
Cross VR, Hester RK, Waugh JS, A model of cell inactivation thechnique,
Rev. Sci Instrum, 47:1486, 1976
Danenberg PV, Thymidylate synthetase-A target enzyme in cancer
chemotherapy, Biochem Biophys Acta 473:73-92, 1977
Danenberg PV, Lockshim A, Thymidylate synthetase - Substrate complex
formation, Mol Cell Biochem 43:49-57, 1978
Danenberg PV, Danenberg KD, Effect of 5,10-methylene tetrahydrofolate
on the dissociation of 5-fluoro-2’-deoxyuridylate from thymidylate
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
162
synthetase evidence for an ordered mechanism, Biochem 17:4018-4024,
1982
De Bruijn EA, Driessen O, Van den Bosch N, et al., A gas-chromatographic
assay for the determination of 5, 6-dihydrofluorouracil and 5-fluorouracil
in human plasma, J Chromatogfr 278:283-289,1983
Decorps M, Blondet P, Reutenauer H, Albrand JP, Remy C, Comparison of
the membrane transport of drug, J Magn Reson 65:100, 1985
Denekamp J, Angiogenesis, neovascular proliferation and vascular
pathophysiology as targets for cancer therapy, Br J Radiol 66:181-196,
1993
Diasio RB, Harris BE, Clinical pharmacology of 5-fluorouracil, Clin
Pharcolinet, 16:215-237,1989
Doty FD, Inners RR, Ellis PD, Comparisons between two monoclonal
antibodies that bind to the same antigen, J Magn Reson 43:399, 1981
El Sayed YM and Sadee W, pharmacokinetics of Anticancer Agents in
Humans, Amsterdam, Elsevier p209-227,1983
Eleff SM, Schnall MD, Ligetti L, Osbakken M, Subramanian VH, Chance B ,
Leigh JS, Jr, Magn Reson Med, 7:412, 1988
Endrich B , Reinhold HS, Gross JF and Intaglietta M , Tissue perfusion
inhomogeneity during early tumor growth in rats, J Natl, Cancer Inst
62:387-395,1979
Engelbrecht C, Lungquist I, Lewan L, et al., Modulation of 5_fluorouracil
metabolism by thymidine in vivo and in vitro studies in RNA-directed
effects in rat liver and rat hepatoma, Biochem Pharmacol, 33:745-750, 1985
Ensminger WD, Rosowsky A, Raso V, et al, A clinical pharmacologic
evaluation of hepatic arterial infusion of 5-fluoro-2,-deoxyuridine and 5-
fluorouracil, Cancer Res 38:3784-3792,1978
Erlemann R, Reiser M, Peters PE, et al, Musculoskeletal neoplasms: static
and dynamic gadopentetate dimeglumine MR imaging, Radiology 171:767,
1989
Evans RM, Laskin JD, Hakala MT, Effect of excess folates and deoxyinosine
on the activity and site of action of 5-fluorouracil, Can Res 41:3288-3295,
1984
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
163
Evans RM, Laskin JD, Hakala MT, Effect of excess folates and deoxyinosine
on the activity and site of action of 5-fluorouracil, Cancer Res 41:3288-
3295,1981
Farris FF, Dedrick RL and King FG, Cisplatin pharmacokinetics:
applications of a physiological model, Toxicol Lett43:117-137,1988
Fernandes DJ and Cranford SK, Resistance of CCRF-CEM cloned sublines to
fluorodeoxyuridine associated with enhanced phosphatase activities,
Biochem Pharmacol 34:125-132, 1985
Finn C and Sadee W, Determination of 5-fluorouracil plasma levels in rats
and man by isotope dilution-mass fragmentography, Cancer Chemother
Rep 59:279-286,1975
Folknan J and Klagsburn M, Angiogenesis factors, Science, 235, 442-447,
1987
Fox SB, Gatter KC, Bicknell R, Going J, et al., Relationship of endothelial cell
proliferation to tumor vascularity in human breast cancer, Cancer Res
53:4161-4167,1993
Froncisz W, Jesmanowicz A, Kneeland JB, Hyde JS, Doxorubicin resistance:
patient studies, Magn Reson Med 3:590, 1986
Gamelin EC, Danquechin-Dorval EM, Dumesnil YF, Maillart PJ, Goudier MJ,
Burtin PC, et al.„ Relationship between 5-fluorouracil dose intensity and
therapeutic response in patients with advanced colorectal cancer
receiving infusional therapy containing 5FU, Cancer 77:441-451, 1996
Gennaro MC, Aime S, Santucci E, et al, Complexes of
diethlenetriaminepentaacetic acid as contrast agents in NMR image:
computer simulation of equilibria in human blood plasma, Anal Chem Acta
233:85,1990
Goitein M and Schutlheiss TE, Strategies for treating possible tumor
extension: some theoretical considerations, Int J Radiat Oncol Biol Phys
11:1519-1528,1985
Goldstein HA, Kashanian FK, Blumetti RF, et al, Safety assessment of
gadopentetate dimeglumine in US clinical trials, Radiology 174:17, 1990
Gore JC, Contrast agents and relaxation effects, In Atlas SW, ed, Magnetic
Resonance Imaging of the brain and spine, New York, Raven Press, 1991
Goren D, Horowitz AT, Zalipsky S, Woodle MC, Yarden Y and Gabizon A,
Targeting of stealth liposomes to erbB-2 receptor: in vitro and in vivo
studies, Br J Cancer 74:1749-1756,1996
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
164
Graeber TG, Osmanian C, Jacks T, Housman DE, Koch CJ, et al., Hypoxia-
mediated selection of cells with diminished apoptotic potential in solid
tumours, Nature 379:88-91,1996
Gray LH, Conger AD, Ebert M, Hornsey S and Scott OC, Concentration of
oxygen dissolved in tissues at the time of irradiation as a factor in
radiotherapy, Br J Radiol 26:638-648,1953
Grist TM, Jesmanowicz A, Kneeland B, Froncisz W, Hyde JS, A Critical
evaluation of the intra-arterrial infusions, Magn Reson Med, 6:253, 1988
Grist TM, Hyde JS, First results of a clinical pilet study on intraarterial
chemotherapy with haemofiltration, J Magn Reson 61:571, 1985
Guerquin-Kern JL, Leteurtre F, Croisy A and Lhoste JM, pH dependence of
5-fluorouracil uptake observed by in vivo 3 1 P and 1 9 F nuclear magnetic
resonance spectroscopy, Cancer Res 51:5770-5773, 1991
Gyves J, Pharmacology of intraperitoneal infusion of 5-fluorouracil and
mitomycin C, Semin Oncol 12:29_32,1985
Halaka MT, Basic for the effect of leucovorin on the activity and site of
action of 5-flurouracil in human cells In Brucker and Rustum (Eds), The
current status of 5-fluorouracil and leucovorin combination, John Willey
and son, New York, p33-39, 1984
Hassenbusch SJ, Anderson JH and Colvin OM, Predicted and actual BCNU
concentrations in normal rabbit brain during intraarterial and
intravenous infusions, J Neuroncol 30:7-18,1996
Heidelberger C, Fluorinated pyrimidines and their nucleosides, In Sortelli
and Johns (Eds) Antineoplastic and immunosuppressive agents, Part II,
pl93-231,Springer-Verlag, New York, 1975
Hemingway DM, Angerson WJ, Anderson JH, Goldberg JA, Me Ardle CS and
Cooke TG, Monitoring blood flow to colorectal liver metastases using laser
doppler flowmetry: the effect of angiotension II, Br J Cancer 66:958-960,
1992
Heywang-Kobrunner SH, Contrast enhanced magnetic resonance imaging
of the breast, Invest Radiol 29:94, 1994
Hirst DG, Denekamp J and Hobson B, Proliferation kinetics of endothelial
and tumor cells in three mouse mammary carcinomas, Cell Tissue Kinet
15:251-261,1982
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
165
Ho DH, Townsend L, Luna MA, et al., Distribution and inhibition of
dihydrouracil dehydrogenase activities in human tissues using 5-
fluorouracil as a substrate, Anticancer Res 6:781-784, 1986
Houghton JA, Houghton PJ, 5-fluorouracil in combination with
hypoxanthine and allopurinol: toxicity and metabolism in xenografts of
human colonic carcinomas in mice, Biochem Pharmacol 29:2077-2080, 1980
Houghton JA, Houghton PJ, Elucidation of pathways of 5-fluorouracil
metabolism in xenografts of human colorectal adenocarcinoma, Eur J Clin
Oncol 19:807-815,1983
Houghton JA, Weiss KD, Williams LG et al., Relationship between 5-fluoro-
5’-deoxy uridyl ate, 2 ’-deoxy uridylate and thy midy late synthase activity
subsequent to 5-fluorouracil administration, in xenografts of human colon
adnocarcinomas, Biochem Pharmacol 35:1351-1358, 1986
Hoult DI, Richards RE, Therapeutic effect of 5-fluorouracil, J Magn Reson
24:71,1979
Hricak H, Bush WH Jr, Cohan R, et al., Genitourinary radiology, Radiology
194:596,1995
Hricak H and Kim B , Contrast enhanced MR imaging of the female pelvis, J
Magn Reson Imaging 3:297, 1993
Hyde Js, Froncisz W. Jesmanowicz A, Kneeland JB, Med Phys 13:1, 1986
Hyde JS. Jesmanowicz A, Grist TM, Froncisz W, Kneeland JB, Magn Reson
Med4:179,1987
Hyde JS, Kneeland JB, in: Wehrli FW, Shaw d, Kneeland B (eds) Biomagnetic
reson imag Verlag Chemie, New York, 1988
Ingraham HA, Tseng B Y , Goulian M, Mechanism of exclusion of 5-
fluorouracil from DNA, Can Res 40:998-1001,1980
Jain RK, Transport of molecules in the tumor interstitium: A review,
Cancer Res 47:3039-3051,1987
Jain RK, Determinants of tumor blood flow: A review, Cancer Res 48:2641-
2658,1988
Jain RK, Delivery of novel therapeutic agents in tumors: physiological
barriers and strategies, J Natl Cancer Inst 81:570-576, 1989
Jain RK, Barriers to drug delivery in solid tumors, Sci Am 271:58-65, 1994
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
166
Jain RK, Whitaker Lecture: delivery of molecules, particles and cells to
solid tumors, Ann Biomed Eng 24:457-473,1995
Jain RK, Delivery of molecular medicine to solid tumors, Science 271:1079-
1080,1996
Jain RK, Delivery of molecular and cellular medicine to solid tumors,
Microcirculation 4:1-23, 1997
Jain, R, Delivery of molecular and cellular medicine to solid tumors,
Journal of Controlled Release v.53, 49-67, 1998
Juliano RL and Stamp D, Pharmacokinetics of liposome-encapsulated anti
tumor drugs, Biochem Pharmacol 27:21-27, 1978
Kameyama M, Fukuda I, Imaoka S, et al., Stagnation of liver blood and
effectiveness of intra-arterial chemotherapy in metastatic colorectal
cancer, Japan J of Cancer and Chemotherapy 16:2894-2896, 1989
Kan S, Fan M, Courtieu J, Rev Sci Instrum 51:887,1980
Kanamaru R, Kakuta H, Sata T, The inhibitory effects of 5-fluorouracil on
the metabolism of peribosomal and ribosomal RNA in L1210 cells in vitro,
Can Chemother Pharmacol 17:43-46,1986
Kent RJ, Heidelberger C, The reduction of fluoruridine monophosphate by
ribonucleotide reductase, Mol Pharmacol 8:465-475, 1972
Kerslake RW, Fox JN, Carle ton PJ, et al., Dynamic contrast enhanced and fat
suppressed magnetic resonance imaging in suspected recurrent
carcinoma of the breast: preliminary experience, J Radiology 67:1158, 1994
Kessel D, Deacon J, Coffey B, Some properties of a pyrimidine
phosphoribosyl transferase from murine leukemic cells, Mol Pharmacol
8:731-739,1972
Kessel D, Cell surface alterations associated with exposure of leukemia
L1210 cells to fluorouracil, Can Res 40:322-324,1980
Kessel D, Hall TC, Wodinsky I, Nucleotide formation as a determinant of 5-
fluorouracil response in mouse leukemias, Science 145:911-913, 1966
Kirsch JE, Basic principles of magnetic resonance contrast agents, Top
Magn Reson Imaging 3:1-3, 1991
Koening H and Patel A, Biochemical basis for fluorouracil neurotoxicity,
Arch Neurol 23:155-160,1970
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
167
Kok RM, De Jong APJM, Van Groeningen CJ, et al, Highly sensitive
determination of 5-fluorouracil in human plasma by capillary gas
chromatography and negative ion chemical ionization mass spectrometry,
J Chromatogr 343:59-66,1985
Kufe DW, Maor PP, Egan EM, Loh E, 5-fluoro-2’-deoxyuridine incorporation
into LI210 DNA, J BiolChem 256:8885-8888,1981
Lamer S, Sigal R, Lassau N, Bosq J, Frouin F, Di Paola M, Mamelle G, Leclere
J, Bittoun J and Di Paola R, Radiologic assessment of intranodal vascularity
in head and neck squamous cell carcinoma. Correlation with histologic
vascular density, Invest Radiol 31:673-679, 1996
Laniado M Kornmesser W, Hamm B, et al., MR imaging of the
gastrointestinal tract: value of Gd-DTPA, A, J Roemtgemp 150:817,1988
Lauffer RB, Principles of MR imaging contrast agents. In Edelman RR, and
Hesselink JR, eds, Clinical magnetic resonance imaging, Philadelphia,
Saunders, 1990
Laufman LR, Kreczowski KA, Roach R, et al., Leucovorin plus 5-
fluorouracil: An effective treatment for metastatic colon cancer, J Clin
0ncol5:1394-1400,1987
Less JR, Posner MC, Boucher Y , Borochovitz D, Wolmark N and Jain RK,
Interstitial hypertension in human breast and colorectal tumors, Cancer
Res 52:6371-6374,1992
Leyva A, van Groeningen CJ, Kraal I, et al., Phase I and pharmacokinetic
studies of high-dose uridine intended for rescue from 5_fluorouracil
toxicity, Cancer Res 44:5928-5933,1984
Maguire JH, Dudley KH, Partial purification and characterization of
dihydropyrimidinase from calf and rat liver, Drug Metab and Dispo 6:601-
610,1978
Malet-Martino MC, Martino R, Lopez A, et al, New approach to metabolism
of 5’deoxy-5-fluorouridine in humans with fluorine-19 NMR, Cancer
Chemother Pharmacol 13:31-35, 1985
Malet-Martino MC, Martino R, Bernadou J, et al., Doxifluridine mainly
excreted in human bile as N-cholyl-_-fluoro-_-alanine, a novel metabolite,
Prec Am Assoc Cancer Res 28:193,1987
Masson E and Zamboni WC, pharmacokinetic optimization of cancer
chemotherapy. Effect on outcomes, Clin Pharmacokinet 32:324-343, 1997
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
168
Meyers CE, The pharmacology of fluoropyrimidines, Pharmacol Rev 33:1-
15,1981
Moses MA, The role of vascularization in tumor metastasis in
“Microcirculation in cancer metastasis” p257-276, CRC Press, 1991
Muller M, Mader RM, Steiner B, et al.„ 5-fluorouracil kinetics in the
interstitial tum or space: clinical response in breast cancer patients,
Cancer Res 57:2598-2601,1997
Murakami M, Yoden E, Kono K, Okamoto Y , et al., Experience in treatment of
patients with locally advanced or recurrent breast cancer: intraarterial
infusion chemotherapy combined with radiotherapy, Acta Radiologica
57:189-194,1997
Murphy-Boesch J, Koretsky AP, Localization of the small tumors, J Maga
Reson 54:526,1983
Nagel TL, Stolk JA, Alderman DW, Schoenborn R, Schweizer MP, Magn
Reson Med 12:271,1990
Naguib FN, El Kouni M, Enzymes of uracil catabolism in normal and
neoplastic human tissue, Can Res 45:5405-5412,1985
Nelson KL and Runge VM, Basic principles. In Runge VM, ed: Enhanced
magnetic resonance imaging, St. Louis, Mosby, 1989
Neuerburg JM, Bohndorf K, Sohn M, et al., Urinary bladder neoplasms:
evaluation with contrast enhanced MR imaging, Radiology 173:739, 1989
Paget S, The distribution of secondary growth in cancer of the breast,
Lancet 1:571,1889
Peters GJ, Kraal I, Laurensse E, et al., Separation of 5-fluorouracil and
uracil by ion-pair reversed-phase high performance liquid
chromatography on a column with porous polymer packing, J Chromatogr
307:464-468,1984
Phillips TA, Howell A, Grieve RJ, et al., Pharmacokinetics of oral and
intravenous fluorouracil in humans, J PharmSci 69:1428-1431,1980
Piscitelli SC, Rodvold KA, Rushing DA and Tewksbury DA, Clin Pharmacol
Ther 53:555-561,1993
Presant CA, Wolf W, Albright MJ, Servis KL et al., Human tumor
fluorouracil trapping: clinical correlation of in vivo 1 9 F nuclear magnetic
resonance spectroscopy pharmacokinetics, J Clin Oncol 8:1868-1873, 1990
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
169
Priest DG, Ledford B E, Doig MT, Increased thymidylate synthetase in 5-
fluorodeoxyuridine resistant cultured hepatoma cells, Biochem Pharmacol
29:1549-1553,1980
Redley RB, Boden J, Deep PA, Harwood PJ, Green AJ and Rogers GT,
Relationship between tumor size and uptake of radiolabeled anti-CEA in a
colon tumor xenograft, Eur J Nucl Med 13:197-202,1987
Reo NV, Ewy CS, Siegfried BA, Ackerman JJH, Regional lymphatic drug
exposure following intraperitoneal administration of 5FU, J Magn Reson
58:76,1984
Reyes P, Hall TC, Synthesis of 5-fluorouridine 5’-phosphate by a
pyrimidine transferase of mammalian origin. Correlation between the
tumor levels of the enzyme and 5-fl.uorouracil-promoted increase in
survival of tumor bearing mice, Biochem Pharmacol 18:2587-2590, 1969
Reyes P and Hall TC, Synthesis of 5-flurouridine-5’-phosphate by a
pyrimidine phosphoribosyl transferase of mammalian origin: Some
properties of the enzyme from P-1534J mouse leukemia cells, Biochemistry
8:2057-2062,1969
Romeo F, Hoult DI, Magnetic resonance imaging of solid tumor patients,
Magn Reson Med 1:44,1984
Rougier P, Ducreux M, Elias D and Lasser P, Hepatic metastases of colorectal
cancers: intra-arterial chemotherapy. Value and indications, Press Med
25:1683-1688,1996
Sadee W and Wong CG, Pharmacokinetics of 5-fluorouracil:
interrelationship with biochemical kinetics in monitoring therapy, Clin
Pharmacokinet 2:437-450,1977
Saini S, Frankel RB, Stark DD, et al., Magnetism: a primer and review, Am J
Roentgenol 150:735,1988
Sands H, Jones PL, Shah SA, Palme D, Vessella RL and Gallagher BM,
Correlation of vascular permeability and blood flow with monoclonal
antibody uptake by human clouser and renal cell xenografts, Cancer Res
48:188-193,1988
Santi DV, Me Henry CS, 5-fluoro-2’-deoxyuridylate: Covalent complex with
thymidylate synthetase, Proc Natl Acad Sci, USA 69:1855-1857,1972
Santi DV, Me Henry CS, Sommer A, Mechanisms of interaction of
thymidylate synthetase with 5-fluorodeoxy uridylate, Biochem 13:471-480,
1974
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
170
Santi DV, Hardy LW, Catalytic mechanism and inhibition of tRNA (Uracil-
5)-methyltransferase: Evidence for covalent catalysis, Biochemistry
26:8599-8606,1987
Sato H, Sugiyama K, Hoshi M, Urushiyama M and Ishizuka K, Angiotensin II
(All) induced hypertension chemotherapy for unresectable gastric
cancer: with reference to resection after down staging, Wold J Surg 19:836-
842,1995
Sawyer RC, Stolfi RL, Martin DS, Incorporation of 5-fluorouracil into
murine bone marrow DNA in vivo,CanRes44:1847-1851,1984
Schenck JF, Hart HR, Jr, Foster TH, Ede;steom WA, Hussain MA, In: Kressel
HY (ed) Magnetic Resonance Annual, Raven Press, New York, p!23, 1986
Scheutz JD, Wallace HJ, Diasio RB, 5-fluorouracil incorporation into DNA of
CF-1 mouse bone marrow cells as a possible mechanism of toxicity, Can Res
44:1358-1363,1984
Scheutz PM, Moer RD, Hyde CM, Role of uridine phosphorylase in the
anabolism of 5-fluorouracil, Biochem Pharmacol 34:3585-3589, 1985
Schnall MD, Subramanian VH, Leigh JS, Pharmacokinetic problems in
peritoneal drug administration, J Magn Reson 67:129, 1986
Schwartz PM, Turek PJ, Hyde CM, et al., Altered plasma kinetics of 5FU at
high dosage in rat and man. Cancer Treat Rep 69:133-136, 1985
Semelka RC, Hricak H, Stevens SK, et al., Combined gadolinium enhanced
and fat saturation MR imaging of renal masses, Radiology 178:803, 1991
Sgouros G, Radioimmunotherapy of micrometastases: sidestepping the
solid-tumor hurdle, J Nucl Med 36:1910-1912,1995
Shani J and Danenberg PV, Evidence that intracellular synthesis of 5-
fluorouridine is compartmentalized, Biochem Biophys Res Commun
122:439-445,1984
Simonetti RG, Liberati A, Angiolini C and Pagliaro L, Treatment of
hepatocellular carcinoma: a systematic review of randomized controlled
trials, Ann Oncol 8:117-136,1997
Sobrero AF, Moir RD, Bertino JR, et al., Defective facilitated diffusion of
nuceosides, a primary mechanism of resistance to 5-fluoro-2’-
deoxyuridine in the HCT-8 human carcinoma lines, Cancer Res 45:3155-
3160,1985
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
171
Sommadossi JP, Gewirtz DA, Diasio RB, Aubert C, Cano JP, Rapid catabolism
of 5-fluorouracil in freshly isolated rat hepatocytes as analysed by HPLC, J
Biol Chem 357:8171-8176,1982
Speyer JL, Collins JM, Dedrick RL, et al., Phase I and pharmacological
studies of 5-fluorouracil administered intraperitoneally, Cancer Res
40:567-572,1980
Steel GG, Growth kinetics of tumors: cell population kinetics in relation to
the growth and treatm ent of cancer, Clarendon Press, Oxford, 1977
Stone HB, Minchinton Al, Lemmon M, Menke D Brown JM, Pharmacological
modification of tumor blood flow: lack of correlation between alteration of
mean arterial blld oressure and chanees in tumor perfusion, Int J Radiat
Oncol Biol Phys 22:79-86,1992
Strand SE, Zanzonico P and Johnson TK, Pharmacokinetic modeling, Med
Phys 20:515-527,1993
Strasser JF, Fung LK, Grossman SA and Saltzman WM, Distribution of 1,3-
bis(2-chloroethyl)-l-nitrosourea and tracers in the rabbit brain after
interstitial delivery by biodegradable polymer implants, J Pharmacol Exp
Ther 275:1647-1655,1995
Sung C, Blaney SM, Cole DE, Balis FM and Dedrick RL, A pharmacokinetic
model of topotecan clearance from plasma and cerebrospinal fluid, Cancer
Res54:5118-5122,1992
Swabb EA, Wei J and Gullino PM, Diffusion and convection in normal and
neoplastic tissues, Cancer Res 34:2814-2822,1974
Swan GW, General applications of optimal control theory in cancer
chemotherapy, IMA J Math Appl Med Biol 5:303-316,1988
Sweeny DJ, Barnes S, Heggie GD, et al, Metabolism of 5-fluorouracil to an N-
cholyl-2-fluoro-_-alanine conjugate, Previously unrecognized role for
bile acids in drug conjugation, Prec Natl Acad Sci USA, 84:5439-5443,1987
Tanaka M, Kimura K, Yoshida S, Increased incorporation of 5-
fluorodeoxyuridine into DNA of human T-lymphoblastic cell lines, Gann
75:986-992,1984
Tannock IF, Population kinetics of carcinoma cells, capillary endothelial
cells and fibroblasts in a transplanted mouse mammary tumor, Cancer Res
30:2470-2476,1970
Terasaki T, Sugiyama Y , Iga T, Sawada Y and Hanano M, Theoretical
consideration of drug distribution kinetics in a noneliminating organ:
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
172
comparison between a “homogeneous (well-stirred)” model and
“nonhomogeneous (tube)” model, J Pharmacokinet Biopharm 13:265-287,
1985
Thomas GD, Chappel MJ, Dykes PW, et al., Effect of dose, molecular size,
affinity, and protein binding on tumor uptake of antibody or ligand: a
biomathematical model, Cancer Res 49:3290-3296,1989
Ulman B , Lee M, Martin DW Jr, Cytotoxicity of 5-fluoro-2’-deoxyuridine:
requirement for reduced folate cofactors and antagonism with
methotrexate, Pron Natl Acad Sci, USA 75:980-983,1978
Van Greningen CJ, Pinedo HM, Heddes J, et al., Pharmacokinetics of 5-
fluorouracil measured with a sensitive mass spectrometric method in
patients during dos escalation schedule, Cancer Res 39:4528-4532, 1989
Vaughan ATM, Anderson P, Dykes PW, Chapman CE and Bradwell AR,
Limitations to the killing of tumors using radiolabeled antibodies, Br J
Radii 60:567-578,1987
Vaupel P, Kallinowski FK and Okunieff P, Blood flow, oxygen and nutrient
supply, and metabolic microenvironment of human tumors: A review,
Cancer Res 49:6449-6465,1989c
Walliser S, Redman K, Effect of 5-fluorouracil and thymidine on the
transmembrane potential and zeta potential of HeLa cells, Can Res 38:3555-
3559,1978
Wehrli FW, The basic of MR contrast. In Atlas SW. ed, Magnetic resonance
imaging of the brain and spine, New York, Raven Press, 1991
Weinhouse S, The Warburg hypothesis fifty years later, Z Krebsforsh,
87:115-126,1976
WeinsteinJN and Van Osdol W, Early intervenstion in cancer using
monoclonal antibodyies and other biological ligands: m icropharm acology
and the “binding site barrier”, Cancer Res 52:2747-2751, 1992
West GW, Weichselbaum R and Little JB, Limited penetration of
methotrexate into human osteosarcoma spheroids as a proposed model for
solid tumor resistance to adjuvant chemotherapy, Cancer Res 40:2665-3668,
1980
Wiig H, Interstitial fluid pressure in DMBA-induced rat mammary tumors,
Scand J Clin Lab Invest 42:159-164,1982
Williams LE, Duda RB, Proffitt RT, et al., Tumor uptake as a fuction of tumor
mass: a mathematic model, J Nucl Med 29:103-109,1988
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
173
Williams LE, Bares RB, Fass J, et al., Uptake of radiolabeled anti-CEA
antibodies in guman colorectal primary tumors as a function of tumor
mass, Eur J Nucl Med 20:345-347,1993
Wolf W, Presant CA, Servis KL, et al., Tumor trapping of 5-fluorouracil: in
vivo 1 9 F NMR spectroscopic pharmacokinetics in tumor-bearing humans
and rabbits, Proc Natl Acad Sci, USA 87:492-496,1990
Wolf W, Kim H, Moy R, Waluch V, Gaudette M, Yu N, Kanekal S, American
Assoc for Cancer Research, 1995
Wolf W, Imaging can be much more than pretty pictures, Pharm ?Sci
12:1821-1822,1995
Wolf W, Waluck V, Presant CA, Kim HK, Sancho AR, Pharmacokinetic
imaging of 5-fluorouracil in tumors using 1 9 F-NMR spectroscopy, Eurospin
Annual p99-103,1996
Wolf W, Waluch V, Presant CA, Noninvasive 1 9 F-NMRS of 5-fluorouraicl in
pharmacokinetics and pharmacodynamic studies, NMR Biomed 11:380-387,
1998
Wolf W, Direct correspondence, 1999
Woodcock TM, Martin DS, Damin LAM, et al., Combination clinical trials
with thymidine and fluorouracil: A phase I and clinical pharmacological
evaluation, Cancer 45:1135-1143,1980
Woodman PW, Sarrif AM, Heidelberger C, Specificity of Pyrimidine
nucleoside phosphorylase and the phosphorolysis of 5-fluoror-2'-
deoxyuriddine, Can Res 40:507-511,1980
Yin MB, Zakrzewski SF, Hakala MT, Relationship of cellular folate cofactor
pools to the activity of 5-fluorouracil, Mol Pharmacol 23:190-197, 1983
Yoshika A, Tanaka S, Hraoka O, Deoxyribonucleotide triphosphate imbalace
-fluorourodeoxyuridne induced DNA double strand breaks on mouse FM3A
cells and the mechanism of cell death, J Biol Chem 262:8235-8241, 1987
Youngner T, Engelbrecht C, Lewan C, et al., Anabolism versus catabolism of
(5-3 H) uridine and its relationship to ribonucleic acid labeling in mouse
liver after partial hepatectomy, Biochem J 178:1-8,1979
Yuan F, Baxter LT and Jain RK, Pharmacokinetic analysis of two-step
approaches using bifunctional and enzyme-conjugated antibodies, Cancer
Res 51:3119-3130,1991
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
174
Zabel H-J, Bader R, Gehrig J, Lorenz WJ, A distribution of peritoneal-
palsma transport, Radiology 165:857-859,1987
Zlotecki RA, Boucher Y , Lee I, Baxter LT and Jain RK, Effect of angiotensin
II induced hypertension on tumor blood flow and interstitial fluid
pressure, Cancer Res 53:2466-2468,1993
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Studies on the mechanism and regulation of the transport of 5-fluorouracil (5FU) into tumor cells

PDF

The effects of East Asian values, coping styles, social support on physical health among Korean American caregivers

PDF

Localization, sorting, and regulation of PepT1

PDF

Pharmacological regulation of chloride, fluid, and solute transport in the pigmented rabbit conjunctiva

PDF

Drug efflux pumps in rabbit conjunctival epithelial cells

PDF

Structure-function studies on the mammalian intestinal dipeptide transporter hPEPT1

PDF

Oxidation induced changes in conjunctival ion transport: Critical role of conjunctival GSH metabolism and transport in ocular surface health

PDF

Design, synthesis, biological testing and QSAR analysis of new Schiff bases of N-hydroxysemicarbazide as inhibitors of tumor cells

PDF

Effect of lipidization on transport and uptake of peptides across rat alveolar epithelial cell monolayers

PDF

Recombinant arginine deiminase (rADI) and arginine metabolism

PDF

Modulation of macrophage glutathione synthesis by oxidized LDL and the effects of glutathione content on cell-mediated LDL oxidation

PDF

Basolateral transport of nucleosides in rabbit tracheal epithelial cells: An equilibrative nucleoside transporter outcome

PDF

Preferential translation of Drosophila heat shock protein 70 and heat shock protein 90 mRNAs during heat shock

PDF

Synthesis of carboplatin and studies on its biotransformation products

PDF

TfR-mediated oral delivery of protein drugs: Oral delivery of recombinant G -CSF -Tf fusion protein and its spacer optimization

PDF

Uptake and transport of desmopressin and its lipidized derivatives in Caco-2 cells

PDF

High mobility group A2 (HMGA2): Molecular dissections of its functions and regulation

PDF

In-situ DNA synthesis on chip with nozzleless MEMS ejector

PDF

Studies of two factors, osteopontin expression and timing of androgen ablation, in prostate tumor progression

PDF

Physiological perturbations in pharmacokinetic animal studies

Asset Metadata

Creator Kim, Hyun Kwon (author)

Core Title Noninvasive in vivo MRI measurements of tumor vascularization and relative tumor osmotic pressure

Degree Doctor of Philosophy

Degree Program Pharmaceutical Sciences

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag health sciences, oncology,Health Sciences, Pharmacology,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-285185

Unique identifier UC11340163

Identifier 3110952.pdf (filename),usctheses-c16-285185 (legacy record id)

Legacy Identifier 3110952.pdf

Dmrecord 285185

Document Type Dissertation

Rights Kim, Hyun Kwon

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA