Download Correlations between dynamic contrast

JOURNAL OF MAGNETIC RESONANCE IMAGING 25:153–159 (2007)
Original Research
Correlations Between Dynamic Contrast-Enhanced
Magnetic Resonance Imaging–Derived Measures of
Tumor Microvasculature and Interstitial Fluid
Pressure in Patients with Cervical Cancer
Masoom A. Haider, MD,1* Igor Sitartchouk PhD,1 Timothy P.L. Roberts, PhD,1
Anthony Fyles, MD,2 Ali T Hashmi,1 and Michael Milosevic, MD2
Purpose: To correlate permeability (rktrans), extracellular
volume fraction (rve), relative to muscle and initial area
under the enhancement curve (IAUC60m) determined by
dynamic contrast-enhanced magnetic resonance imaging
(DCE-MRI) with in vivo measurements of interstitial fluid
pressure (IFP) in patients with cervical cancer.
Materials and Methods: DCE-MRI and IFP measurements
were performed of cervical tumors of 32 patients prior to
therapy. Median tumor rktrans and rve were derived from a
bidirectional two-compartment model using an input function derived from muscle. Median IAUC60m was defined as
the integral of tumor enhancement in the first 60 seconds
divided by the similar muscle enhancement integral. These
parameters were correlated with the mean tumor IFP.
Results: There was a significant negative correlation between IAUC60m and IFP (r ⫽ – 0.42, P ⫽ 0.016) and between
rktrans and IFP (r ⫽ – 0.47, P ⫽ 0.008). The was no significant
correlation between IFP and rve.
Conclusion: There is a moderate negative correlation between IAUC60m, rktrans, and IFP in cervical cancer. This
suggests that these parameters may be of value in assessment of tumor behavior.
Key Words: permeability; dynamic contrast-enhanced
MRI; cancer of the cervix; interstitial fluid pressure; radiation therapy
J. Magn. Reson. Imaging 2007;25:153–159.
© 2006 Wiley-Liss, Inc.
1
Department of Medical Imaging, University Health Network and Mount
Sinai Hospital, University of Toronto, Canada.
2
Radiation Medicine Program, University Health Network, Princess
Margaret Hospital, Toronto, Canada.
Contract grant sponsor: National Cancer Institute of Canada; Contract
grant sponsor: Terry Fox Run.
*Address reprint requests to: M.A.H., MD, Dept. of Medical Imaging,
Princess Margaret Hospital, 610 University Ave. Toronto, Ontario, Canada, M5G 2M9. E-mail: [email protected]
Received November 23, 2004; Accepted August 31, 2006.
DOI 10.1002/jmri.20795
Published online 15 December 2006 in Wiley InterScience (www.
interscience.wiley.com).
© 2006 Wiley-Liss, Inc.
IT HAS BEEN DEMONSTRATED that high interstitial
fluid pressure (IFP) is correlated with shorter diseasefree survival and a higher likelihood of pelvic recurrence
as well as distant metastases in cervical cancer patients
who are treated with radiotherapy (1). IFP in normal
tissues is close to atmospheric pressure but is elevated
in most solid malignancies, with values ranging from 10
to 100 mm Hg (1–5). IFP is thought to be elevated as a
result of both abnormal tumor microvasculature resulting from unregulated angiogenesis and lack of functional lymphatics (6 – 8). The high vascular permeability
of tumor vessels results in elevated interstitial protein
concentrations and the oncotic pressure gradient
across tumor capillaries approaches zero (9). This leads
to increased leakage of fluid from the vascular to the
interstitial space, where it accumulates and distends
the elastic interstitial matrix, causing IFP to rise. In
most tumors, it is likely that the balance of factors that
determine fluid flux across the vessel wall and through
the interstitium is such that IFP at steady-state is equal
to the average capillary pressure (8,10). In this situation, differences in IFP from tumor to tumor probably
reflect underlying variability in capillary blood flow resistance and capillary pressure. Thus, abnormalities of
the tumor microvasculature likely directly influence
IFP.
Many investigators have used dynamic contrast-enhanced MRI (DCE-MRI) to study tumor angiogenesis
and patient outcome for various tumors including cervical cancer (11–16). This is generally done by quantifying signal changes in images of a tumor obtained at
high temporal resolution after bolus injection of a low
molecular weight intravenous contrast (17). A variety of
descriptive parameters can be derived from the shape of
the tumor enhancement curve, including peak enhancement, initial slope, residual enhancement or initial area under the enhancement curve (IAUC). Of
these, the ratio of IAUC in the tumor to IAUC in muscle
(used as a reference), which is thought to be related to
a combination of tumor blood inflow, bulk perfusion,
and the fraction of tumor interstitial space, is least
affected by alterations in bolus duration and noise
153
154
Haider et al.
Figure 1. Appearance of cervical cancer on T2-weighted and DCE-MRI. Axial T2-weighted image (a) shows a large tumor (black
arrows). At the same axial level, precontrast (b), 8 seconds (c), and 24 seconds (d) postcontrast 3D SPGR image shows ROI drawn
to encompass the tumor (dotted line) and muscle (solid line). The median value from each of these ROIs was used to plot the
relaxivity curves shown in Fig. 2.
(18,19). IAUC is particularly advantageous because of
the simplicity of the analysis. Unlike IAUC, pharmacokinetic modeling can be used to derive parameters that
may relate more directly to the microvasculature such
as permeability-surface area product, perfusion, blood
volume, and extracellular volume fraction (20,21).
Given the relationship of both IFP and DCE-MRI to the
tumor microvasculature, we hypothesized that DCEMRI– derived parameters may be related to IFP in cervical cancer.
To date no studies have been performed studying the
relationship of DCE-MRI parameters with IFP. Such a
relationship could provide insight into in vivo tumor
physiology and treatment response. A relationship
would also be important clinically as IFP has been
shown to be an independent predictor of patient outcome in cervical cancer in patient treated with radiation
therapy.
The purpose of this study was to correlate parameters
derived from DCE-MRI with IFP measurements in patients with cervical cancer. In particular, the initial area
under the enhancement curve relative to muscle
(IAUC60m), vascular permeability relative to muscle
(rktrans), and extracellular volume fraction relative to
muscle (rve) as determined by MRI were chosen for this
comparison because of their robustness and pathophysiologic relationship to elevated IFP.
MATERIALS AND METHODS
Patients with cervical cancer who were candidates for
curative radiation therapy with or without concurrent
cisplatin chemotherapy were eligible for this study. Radiotherapy and concurrent chemotherapy is standard
treatment for locally extensive or lymph node–positive
cervical cancer. Patients were considered suitable for
radiation therapy if they had an International Federation of Gynecology and Obstetrics (FIGO) stage IB tumor greater than 4 cm in maximum size by clinical
examination and/or MRI, a FIGO stage IIA to IV tumor
of any size, or enlarged pelvic or paraaortic lymph
nodes with a short-axis diameter greater than 0.8 cm by
computed tomography (CT) or MRI.
DCE-MRI was performed through the cervical tumors
of 37 patients prior to initiation of treatment between
May 15, 2003 and April 11, 2005. Adequate images
were not obtained in four of these patients due to technical failure in image acquisition or severe patient motion, and DCE-MRI measurements could not be made
in one patient who had a very small tumor, leaving 32
patients for analysis. Patient mean age was 48.6 (range
30.5–72) years. Of the 32 patients, 21 had squamous
carcinoma, eight had adenocarcinoma, and three adenosquamous carcinoma. Tumor stage was based on
FIGO guidelines at examination under anesthesia as
previously reported (1,22). FIGO stage was IB in 12
patients, IIB in 10, IIIB in nine, and IV in one. Mean
tumor diameter for the whole population as measured
with MRI was 4.5 cm (range 2.8 – 8.0 cm). Adenopathy
based on imaging criteria (⬎8 mm in short axis) was
present in 14 of 32 patients. Approval for this study was
obtained from the institutional ethics review board. Informed consent was obtained in all patients.
Dynamic contrast-enhanced MRI (3D spoiled gradient echo [SPGR], fat saturation, TR ⫽ 6.6 msec, TE ⫽
2.2 msec, matrix size ⫽ 256 ⫻ 128, three-quarter rectangular field of view, flip angle ⫽ 40°, temporal resolution ⫽ 8 seconds, slice thickness ⫽ 10 mm, scan time ⫽
400 seconds, and slice locations ⫽ 6) was performed
through the cancer using a 1.5-T MRI system (Excite or
Echospeed; General Electric Medical Systems, Milwaukee, WI, USA) with a four- or eight-channel surface
phased array coil. Imaging was during and following
injection of 0.1 mmol/kg body weight of gadodiamide
(Omniscan; GE Healthcare Ltd., Amersham, UK) via a
power injector at a rate of 2 mL/second. Injection and
scanning were initiated simultaneously.
Cervical tumor was defined by the enhancement pattern and T2-weighted imaging by a radiologist with 10
years of experience in the interpretation of cervical MRI.
A region of interest (ROI) was drawn to encompass the
tumor, avoiding vessels and areas of cavitation on a
single axial slice that showed the largest amount of
tumor. An ROI was also drawn over each gluteal muscle
as far as possible from the skin surface and avoiding
vascular ghost artifact or large vessels (Fig. 1). Median
values of the ROIs were used. Median values were chosen as opposed to generating voxel-by-voxel parametric
maps to help reduce errors from image noise and patient motion. The time-point of onset of tumor enhancement was assessed by visual inspection of the source
dynamic images. The mean tumor size was calculated
for each patient. This was defined as the mean of the
maximal perpendicular axial and longitudinal measurements taken from the T2-weighted images.
The IAUC60m was calculated by integrating the median increase in signal intensity from baseline over the
first 60 seconds after contrast arrival in the tumor, and
DCE-MRI and IFP in Cervical Cancer
155
dividing by the similar integral for the gluteal muscle.
To keep the analysis as simple as possible, no correction to calculate relaxivity from signal intensity was
performed.
To assess vascular permeability relative to muscle
(rktrans) and extracellular volume fraction relative to
muscle (rve), a more complex analysis was performed. It
is assumed that relaxivity (R) is linearly related to the
concentration of contrast agent. To determine relaxivity
it was necessary to convert from MRI signal intensity (S)
to R using Eq. [1] below in two steps. First, S0 (a constant describing the scanner gain and proton density)
was calculated by using the baseline signal intensity of
the gluteal muscle and a previously published mean T1
value of muscle of 856 msec (23) and the known values
of flip angle (␣) and TR from the pulse sequence. Once
S0 was known, the measured signal intensity in both
muscle an tumor during contrast injection (S) could be
converted to R using Eq. [1]. It should be noted that this
method assumes similar S0, which is related to proton
density and signal gain, for muscle and tumor. The
assumption of similar proton density for tumor and
muscle is not unreasonable as the concentration of
hydrogen in soft tissues is similar. However, signal gain
may not always be similar when there are spatially
dependent changes in signal intensity from a phased
array surface coil. To minimize the difference in S0 between muscle and tumor the ROIs for muscle were
drawn as far from the surface coil and as close to the
tumor as possible (Fig. 1).
冋
册
1 ⫺ e⫺TR共R兲
1
S ⫽ S 0 共sin␣兲
, R⫽ .
T1
1 ⫺ e⫺TR共R兲COS ␣
(1)
The change in relaxivity from baseline (Rmuscle(T)) for
muscle was fitted using a biexponential model to help
improve model fitting. Since we know the change in
relaxivity of the tumor during contrast injection (Rtumor(T)) we can fit Eq. [2] and solve for tumor vascular
permeability relative to muscle (rktrans), and extracellular volume relative to muscle (rve) (Eqs. [2] and [3]), as
described by Yankeelov et al (24). Curve fitting was
performed using in-house software developed using IDL
v6.0 (RSI, Boulder, CO, USA). Because the temporal
resolution of the source data was eight seconds, the
analytic method allowed incrementally tested curve fits
with a time shift for each curve of up to eight seconds.
The shifted curve with the best fit (the lowest sum of
squared residuals) was used for analysis. A model fit
could not be obtained in one patient. A typical curve fit
for muscle and tumor is shown in Fig. 2.
T
冕
R tumor 共T兲 ⫽ a 1 关R muscle 共T兲兴 ⫹ a 1 关a 2 ⫺ a 3 兴 e⫺a3共T⫺t兲dt
0
where
(2)
Figure 2. Typical tumor enhancement curve (change in relaxivity vs. time). The muscle curve is of lower amplitude and is
smoothed using a biexponential fit (dotted line). The tumor
curve is of higher amplitude and is fitted using a pharmacokinetic model (Eq. [2]) that uses muscle as a reference tissue.
The rktrans (permeability relative to muscle) in this case was 6.5
and the extracellular volume fraction relative to muscle (rve)
was 2.68. The data points were taken from the images shown
in Fig. 1.
a1 ⫽
k trans,tumor
k trans,muscle
,a ⫽
,a 3
k trans,muscle 2
␯ e,muscle
⫽
k trans,tumor
, rk trans ⫽ a 1, rv e ⫽ a 1(a 2)/a 3
␯ e,tumor
(3)
In vivo measurement of IFP was performed after the
MRI and before initiation of radiation or chemotherapy.
IFP was measured using a wick-in-needle apparatus, as
described in detail elsewhere (1). All of the measurements were made during examination under anesthesia with the patients in the lithotomy position. Anesthesia was administered using intravenous (i.v.) propofol
and inhaled nitrous oxide, and the inspired oxygen concentration was maintained at 40%. Clinical examination and pelvic MRI were used to assure that the IFP
measurements were made in tumor and not inadvertently in adjacent normal tissue. IFP was measured at
one to five spatially separated locations around the
circumference of the visible cervical tumor, at a depth of
approximately 2 cm from the surface. Multiple measurements of IFP are necessary in each tumor to account for regional heterogeneity, as described previously (25,26). The mean of all measurements was used
as a summary measure of tumor IFP.
A check for normality was performed (KolmogorovSmirnov test) and a significant difference from a normal
distribution was not found for the variables being evaluated. Pearson correlation coefficients were used to determine if there was a significant positive or negative
relationship between IFP, tumor size, and the DCE-MRI
measures. Correlation coefficients different from 0 with
a P-value ⬍ 0.05 were considered significant. Correlations between 0.4 and 0.6 were considered moderate
156
Figure 3. IAUC60m in the first 60 seconds vs. IFP. A negative
correlation (r ⫽ – 0.42, P ⫽ 0.016, N ⫽ 32) between tumor IFP
and the IAUC divided by the initial area under the muscle
enhancement curve for the first 60 seconds after contrast
arrival (IAUC60m) is demonstrated.
while correlations from 0.7 to 1.0 were considered
strong (27). A t-test was used to compare groups of
patients. Equal or unequal variances for comparisons
were used when appropriate. For tumor stage and
grade paired comparisons were sequentially performed
using by grouping patients above or below a tumor
stage or grade. SPSS v11.0 software was used for statistical analysis (SPSS Inc., Chicago, IL, USA).
Haider et al.
Figure 4. Relative permeability vs. IFP. There was a significant negative correlation (r ⫽ – 0.47, P ⫽ 0.008, N ⫽ 31) between tumor IFP and the permeability of tumor relative to
muscle (rKtrans).
and is less affected by alterations in bolus duration and
noise (18,19). It is possible that higher values of IFP are
associated with higher geometric and viscous blood
flow resistance, such that a greater proportion of the
systemic blood perfusion pressure is developed across
the capillary bed and reflected in the measured IFP (28).
This situation would also be associated with lower
RESULTS
Median IFP was 16.3 mmHg (range of 3.86 – 40.97
mmHg). IAUC60m had a median of 4.0 (range
2.03–12.77), rktrans 4.65 (range 0.09 –15.94), and rve
2.56 (range 0 –11.99). There was a significant negative
correlation between IAUC60m and IFP (r ⫽ – 0.42, P ⫽
0.016, N ⫽ 32) (Fig. 3) and between rktrans and IFP (r ⫽
– 0.47, P ⫽ 0.008, N ⫽ 31) (Fig. 4). There was no significant correlation between IFP and rve (P ⫽ 0.52) (Fig. 5).
There was a strong significant positive correlation between rktrans and IAUC60m (r ⫽ 0.79, P ⬍ 0.0001) (Fig. 6).
There was no correlation between mean tumor size on
MRI and any of the DCE-MRI parameters (r ⫽ – 0.03 to
0.05, P ⫽ 0.77– 0.88). There was no significant difference in DCE-MRI parameters when grouped by tumor
stage, grade, or lymph node status.
DISCUSSION
IAUC60m is a simple measure that correlated negatively
with tumor IFP in this study. IAUC60m is thought to be
related to a combination of tumor blood inflow, bulk
perfusion, and the fraction of tumor interstitial space
Figure 5. rve vs. IFP. There was no significant correlation
between the rve relative to muscle and IFP. Even with the
outlier removed a significant correlation was not found.
DCE-MRI and IFP in Cervical Cancer
Figure 6. Strong correlation between rktrans and IAUC60m.
There was a strong correlation between rktrans and IAUC60m,
suggesting that these parameters may be providing similar
information.
blood flow and bulk perfusion, and a lower IAUC60m,
consistent with the negative correlation between IFP
and IAUC60m that we observed.
Microvascular permeability relative to muscle (rktrans)
was negatively correlated with IFP. IFP is generally
thought to be elevated in tumors because of high vascular permeability in regions of angiogenesis, low oncotic pressure gradients from protein leakage, and impaired lymphatic drainage from the interstitium (6 – 8).
Therefore, it seems counterintuitive that IFP appears
negatively correlated with permeability. There are a
number of potential explanations for this finding. Perhaps the most likely explanation is that rktrans is not
reflecting permeability alone but a combination of microvasculature features depending on permeability,
capillary surface area, and blood flow. It has been suggested that the geometric and viscous blood flow resistance in tumors may be substantially greater than the
resistance to fluid flow across capillary walls from the
vasculature into the interstitium (29,30), especially in
tumors with high IFP. In these circumstances, transport of gadodiamide into the tumor interstitium, and
therefore the estimates of permeability in this study,
might be more strongly influenced by vascular flow
resistance rather than by vessel permeability. Blood
flow resistance should be lower and more comparable
to transvascular flow resistance in tumors with lower
IFP, and permeability values should be correspondingly
higher. It is well recognized that, when using low molecular weight contrast agents such as gadodiamide,
flow may dominate ktrans estimates in regions where
flow and permeability are comparable (17).
The hypothesis that the permeability estimates in
this study actually reflect blood flow resistance rather
than vascular permeability is perhaps the most likely
157
explanation for the negative correlation between ktrans
and IFP. However, other possibilities should be considered. One might expect that high IFP would restrict
transport of gadodiamide from the vascular to extravascular space. However, given the very small molecular
size of this agent it is unlikely that high IFP, which
mainly limits convective transport of macromolecules
(8), would have a substantial effect on the mostly diffusive transport of such a small molecule. Yet another
hypothesis is that tumors with high IFP have restricted
fluid movement in the extravascular, extracellular
space related to poor lymphatic drainage, and that in
the extremely high permeability environment of a malignancy we are in fact measuring interstitial space permeability and not vascular permeability when using low
molecular weight contrast agents.
In this study we did not find a correlation between rve
and IFP. It is difficult to draw conclusions from this lack
of correlation. One might expect tumors with high IFP
to have distention of the elastic extravascular extracellular space and thus a positive correlation between IFP
and rve. The derivation of rve is based on the assumption of a rapid equilibration of contrast agent between
the intravascular and extravascular, extracellular
space. This may not be true in areas of high IFP in
which protein content and altered lymphatic drainage
could alter the efflux and relaxation times of contrast
agent in the interstitium. Thus rve as derived in the
model used in this study may not reflect the true interstitial volume. It is possible that models that incorporate blood flow and vascular resistance as well as vascular permeability and the extravascular extracellular
space may be required to develop a better understanding of the relationship between IFP and tumor microvasculature.
There are a number of methods of analyzing dynamic
contrast enhanced MRI data. IAUC and ktrans are two
parameters that have been put forward for use in drug
trials (31). The methodologies used here rely on muscle
as a reference tissue (19,24). This was necessary as
there were changes in signal intensity of the source
images related to manufacturer related changes in image reconstruction as well as the use of four- and eightchannel surface arrays. By using muscle, the variability
introduced by these factors as well as bolus duration
could be minimized. The use of muscle as a reference
tissue has been advocated by others, as a true vascular
input function from the arterial circulation to the tumor
is difficult to obtain (24,32). Bolus duration differed
significantly in this study from patient to patient as
contrast was injected at a fixed rate but the volume
varied dependent on patient body weight. The use of
muscle as a reference tissue for IAUC measures would
be expected to reduce the effect of bolus duration on the
IAUC measure (19).
There was a strong correlation between IAUC60m and
rktrans and they may be interchangeable in the context
of this study. IAUC60m is an attractive measure as it is
easy to calculate and relatively noise insensitive. Rapid
calculation of tumor parametric maps that would allow
for assessment of tumor heterogeneity not considered
in the paper would also be possible with only 60 seconds of data acquisition after contrast bolus arrival.
158
There is only one other article that has studied the
relationship between dynamic contrast-enhancement
and IFP in cervical cancer (33). In that work, no relationship between IFP and perfusion parameters was
found but CT was used as opposed to MRI. CT has lower
contrast sensitivity. In addition, a different analytic
method was used for CT. Both these factors could explain why similar relationships were not found between
CT and MRI methodologies.
There are some limitations to this study. We have
only shown a moderate correlation between IFP and
tracer kinetic modeling derived parameters. This is not
surprising. IFP is affected by a number of pathophysiologic processes beyond simply microvascular blood
flow, permeability, and the size of the interstitial space.
In particular, the relationship between the effect of altered lymphatic drainage is not accounted for in the
typical pharmacokinetic models used with DCE-MRI,
nor is the heterogeneity of the tumor itself or its microvascular architecture. Temporal and spatial variations
in blood flow are known to occur and may affect measurements taken on different days. The assumed linearity between relaxivity and gadolinium concentration
relies on an assumption of unimpeded access of water
molecules in the remote tissue compartment to the
tracer. This may not always be the case and it is unclear
to what degree this may have affected the results. In
addition, we were unable to precisely identify the locations where IFP was measured within the tumors. Precise anatomic localization would require that the IFP
measurements were performed under direct MRI guidance as cervical cancer is poorly visualized by other
conventional imaging modalities such as ultrasound
and CT. The reliance on a summary measure of ROI
would have introduced some variability in measurement related to tumor heterogeneity that is unavoidable
with this study design. Given all of these factors one
could argue that the presence of a moderate correlation
is in fact very good evidence of an important role of
tumor microvasculature in the pathophysiologic process that leads to elevated IFP.
Although rktrans and IAUC60m cannot be used as surrogate markers of IFP they may be important predictors
of outcome because of their relationship to IFP. The
lack of any relationship between tumor size, grade, or
stage and DCE-MRI parameters raises the possibility
that one of these parameters could be an independent
outcome predictor. Continued data collection is under
way to determine their potential clinical value. The use
of the IAUC60m measure may be particularly useful as it
is quite easy to perform and is robust and relatively
insensitive to noise.
In conclusion, there is a moderate negative correlation between the initial area under the enhancement
curve relative to muscle (IAUC60m), permeability relative
to muscle (rktrans), and the IFP in cervical cancer. These
DCE-MRI parameters may be of value in assessment of
tumor behavior.
Haider et al.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
REFERENCES
1. Milosevic M, Fyles A, Hedley D, et al. Interstitial fluid pressure
predicts survival in patients with cervix cancer independent of
23.
clinical prognostic factors and tumor oxygen measurements. Cancer Res 2001;61:6400 – 6405.
Less JR, Posner MC, Boucher Y, Borochovitz D, Wolmark N, Jain
RK. Interstitial hypertension in human breast and colorectal tumors. Cancer Res 1992;52:6371– 6374.
Curti BD, Urba WJ, Alvord WG, et al. Interstitial pressure of subcutaneous nodules in melanoma and lymphoma patients: changes
during treatment. Cancer Res 1993;53:2204 –2207.
Gutmann R, Leunig M, Feyh J, et al. Interstitial hypertension in
head and neck tumors in patients: correlation with tumor size.
Cancer Res 1992;52:1993–1995.
Nathanson SD, Nelson L. Interstitial fluid pressure in breast cancer, benign breast conditions, and breast parenchyma. Ann Surg
Oncol 1994;1:333–338.
Leu AJ, Berk DA, Lymboussaki A, Alitalo K, Jain RK. Absence of
functional lymphatics within a murine sarcoma: a molecular and
functional evaluation. Cancer Res 2000;60:4324 – 4327.
Padera TP, Kadambi A, di Tomaso E, et al. Lymphatic metastasis in
the absence of functional intratumor lymphatics. Science 2002;
296:1883–1886.
Baxter L, Jain R. Transport of fluid and macromolecules in tumors
I. Role of interstitial pressure and convection. Microvasc Res 1989;
37:77–104.
Jain RK. Transport of molecules in the tumor interstitium: a review.
Cancer Res 1987;47:3039 –3051.
Boucher Y, Baxter LT, Jain RK. Interstitial pressure gradients in
tissue isolated and subcutaneous tissues: implications for therapy.
Cancer Res 1990;50:4478 – 4484.
Loncaster JA, Carrington BM, Sykes JR, et al. Prediction of
radiotherapy outcome using dynamic contrast enhanced MRI of
carcinoma of the cervix. Int J Radiat Oncol Biol Phys 2002;54:
759 –767.
Hawighorst H, Weikel W, Knapstein PG, et al. Angiogenic activity of
cervical carcinoma: assessment by functional magnetic resonance
imaging-based parameters and a histomorphological approach in
correlation with disease outcome. Clin Cancer Res 1998;4:2305–
2312.
Hawighorst H, Knapstein PG, Knopp MV, Vaupel P, van Kaick G.
Cervical carcinoma: standard and pharmacokinetic analysis of
time-intensity curves for assessment of tumor angiogenesis and
patient survival. MAGMA 1999;8:55– 62.
Yamashita Y, Baba T, Baba Y, et al. Dynamic contrast-enhanced
MR imaging of uterine cervical cancer: pharmacokinetic analysis
with histopathologic correlation and its importance in predicting the outcome of radiation therapy. Radiology 2000;216:803–
809.
Mayr NA, Hawighorst H, Yuh WT, Essig M, Magnotta VA, Knopp
MV. MR microcirculation assessment in cervical cancer: correlations with histomorphological tumor markers and clinical outcome. J Magn Reson Imaging 1999;10:267–276.
Mayr NA, Yuh WT, Zheng J, et al. Prediction of tumor control in
patients with cervical cancer: analysis of combined volume and
dynamic enhancement pattern by MR imaging. AJR Am J Roentgenol 1998;170:177–182.
Padhani AR, Dzik-Jurasz A. Perfusion MR imaging of extracranial
tumor angiogenesis. Top Magn Reson Imaging 2004;15:41–57.
Evelhoch JL, LoRusso PM, He Z, et al. Magnetic resonance imaging
measurements of the response of murine and human tumors to the
vascular-targeting agent ZD6126. Clin Cancer Res 2004;10:3650 –
3657.
Evelhoch JL. Key factors in the acquisition of contrast kinetic data
for oncology. J Magn Reson Imaging 1999;10:254 –259.
Tofts PS, Kermode AG. Measurement of the blood-brain barrier
permeability and leakage space using dynamic MR imaging. 1.
Fundamental concepts. Magn Reson Med 1991;17:357–367.
Tofts PS, Brix G, Buckley DL, et al. Estimating kinetic parameters
from dynamic contrast-enhanced T(1)-weighted MRI of a diffusable
tracer: standardized quantities and symbols. J Magn Reson Imaging 1999;10:223–232.
Fyles A, Milosevic M, Hedley D, et al. Tumor hypoxia has independent predictor impact only in patients with node-negative cervix
cancer. J Clin Oncol 2002;20:680 – 687.
de Bazelaire CM, Duhamel GD, Rofsky NM, Alsop DC. MR imaging
relaxation times of abdominal and pelvic tissues measured in vivo
at 3.0 T: preliminary results. Radiology 2004;230:652– 659.
DCE-MRI and IFP in Cervical Cancer
24. Yankeelov TE, Luci JJ, Lepage M, et al. Quantitative pharmacokinetic analysis of DCE-MRI data without an arterial input function:
a reference region model. Magn Reson Imaging 2005;23:519 –529.
25. Milosevic MF, Fyles AW, Wong R, et al. Interstitial fluid pressure in
cervical carcinoma: within tumor heterogeneity, and relation to
oxygen tension. Cancer 1998;82:2418 –2426.
26. Pitson G, Fyles A, Milosevic M, Wylie J, Pintilie M, Hill R. Tumor size
and oxygenation are independent predictors of nodal diseases in patients with cervix cancer. Int J Radiat Oncol Biol Phys 2001;51:699–703.
27. Zou KH, Tuncali K, Silverman SG. Correlation and simple linear
regression. Radiology 2003;227:617– 622.
28. Zlotecki RA, Baxter LT, Boucher Y, Jain RK. Pharmacologic modification of tumor blood flow and interstitial fluid pressure in a
human tumor xenograft: network analysis and mechanistic interpretation. Microvasc Res 1995;50:429 – 443.
29. Netti PA, Roberge S, Boucher Y, Baxter LT, Jain RK. Effect of
transvascular fluid exchange on pressure-flow relationship in tu-
159
30.
31.
32.
33.
mors: a proposed mechanism for tumor blood flow heterogeneity.
Microvasc Res 1996;52:27– 46.
Baish JW, Netti PA, Jain RK. Transmural coupling of fluid flow in
microcirculatory network and interstitium in tumors. Microvasc
Res 1997;53:128 –141.
Leach MO, Brindle KM, Evelhoch JL, et al. Assessment of antiangiogenic and antivascular therapeutics using MRI: recommendations for appropriate methodology for clinical trials. Br J Radiol
2003;76(Spec No 1):S87–S91.
Kovar DA, Lewis M, Karczmar GS. A new method for imaging
perfusion and contrast extraction fraction: input functions derived from reference tissues. J Magn Reson Imaging 1998;8:
1126 –1134.
Haider MA, Milosevic M, Fyles A, et al. Assessment of the tumor
microenvironment in cervix cancer using dynamic contrast enhanced CT, interstitial fluid pressure and oxygen measurements.
Int J Radiat Oncol Biol Phys 2005;62:1100 –1107.