CT Imaging • Original Research

Sabir et al.
Perfusion MDCT of Antiangiogenic Therapy Response

CT Imaging
Original Research

Perfusion MDCT Enables Early

Adeel Sabir 1,2
Rachel Schor-Bardach1,2
Carol J. Wilcox 2,3
Syed Rahmanuddin1,2
Michael B. Atkins 2,4
Jonathan B. Kruskal2,3
Sabina Signoretti2,5
Vassilios D. Raptopoulos 2,3
S. Nahum Goldberg1,2,3
Sabir A, Schor-Bardach R, Wilcox CJ, et al.
Keywords: antiangiogenic therapy, MDCT, perfusion CT,
perfusion imaging, renal cell carcinoma, sorafenib
DOI:10.2214/AJR.07.2848
Received July 10, 2007; accepted after revision
January 10, 2008.
This study was supported by a grant from the National
Cancer Institute (NCI) Dana Farber/Harvard Cancer
Center (DF/HCC) Renal Cancer SPORE grant (no. 1 P50
CA10194-01). Sorafenib (BAY 43-9006) used for this study
was provided by Bayer, Inc., West Haven, CT.
R. Schor-Bardach is a recipient of a fellowship grant from
The American Physicians Fellowship for Medicine in Israel.
1
Minimally Invasive Tumor Therapy Laboratory,
Beth Israel Deaconess Medical Center, Harvard Medical
School, Boston, MA.
Detection of Therapeutic Response
to Antiangiogenic Therapy
OBJECTIVE. The objective of our study was to determine whether perfusion CT can be
used to detect early changes in therapeutic response to antiangiogenic therapy in an animal
tumor model.
MATERIALS and methods. Twenty-five rats implanted with R3230 mammary ade-
nocarcinoma (diameter, 1.2–2.0 cm) randomly received 7.5 or 30 mg/kg of an antiangiogenic
agent, sorafenib, by daily gavage for 4 (n = 4), 9 (n = 9), or 14 (n = 5) days. Seven untreated animals
served as a control group. Perfusion MDCT was performed at days 0, 4, 9, and 14 with 0.4 mL of
ioversol (350 mg/mL) and included four 5-mm slices covering the entire tumor volume. Changes
in tumor growth were determined by volumetric analysis of CT data. Serial changes in tumor
volume and blood flow were assessed and correlated with pathology findings.
RESULTS. All control tumors grew larger (from 2.0 ± 0.7 cm3 at day 0 to 5.9 ± 1.0 cm3 at day
14), whereas all treated tumors shrank (from 2.5 ± 1.1 to 2.1 ± 1.0 cm3), with a statistically sig-
nificant rate of growth or shrinkage in both groups (p < 0.05). Although perfusion in the control
tumors changed little from day 0 to day 14 (day 0, 18.1 ± 9.2 mL/min/100 g; day 4, 15.8 ± 5.6; day
9, 21.7 ± 12.2; day 14, 27.7 ± 34), in the sorafenib group, the mean blood flow was significantly
lower at day 4 (5.2 ± 3.2 mL/min/100 g, 77% decrease), day 9 (6.4 ± 4.0 mL/min/100 g, 66% de-
crease), and day 14 (6.3 ± 5.2 mL/min/100 g, 83% decrease) compared with day 0 (23.8 ± 11.6 mL/
min/100 g) (p < 0.05). Poor correlation was seen between changes in blood flow and tumor volume
for days 0–9 (r2 = 0.34), 4–9 (r2 = 0.0004), and 9–14 (r2 = 0.16). However, when comparing day 4
images with days 9 and 14 images, seven of 14 (50%) sorafenib-treated tumors had focal areas of
new perfusion that correlated with areas of histopathologic viability despite the fact that these tu-
mors were shrinking in size from day 4 onward (day 4, 2.18 ± 0.8 cm3; day 9, 1.98 ± 0.8 cm3).
CONCLUSION. Perfusion MDCT can detect focal blood flow changes even when the tumor
is shrinking, possibly indicating early reversal of tumor responsiveness to antiangiogenic therapy.
Given that changes in tumor volume after antiangiogenic therapy do not necessarily correlate with
true treatment response, physiologic imaging of tumor perfusion may be necessary.

T
he treatment of many vascular tu- [6]. Sorafenib has been shown to produce tu-
2
Renal Cancer Program, Dana Farber/Harvard Cancer mors and of renal cell carcinoma mor shrinkage in up to 80% of patients with
Center, Boston, MA.
(RCC) in particular has undergone advanced RCC and to significantly extend me-
3
Department of Radiology, Beth Israel Deaconess a radical change over the past few dian progression-free survival. Despite these
Medical Center, 1 Deaconess Rd., WCC 308-B, Boston, years because of the advent of targeted antian- impressive results, sorafenib does not produce
MA 02215. Address correspondence to S. N. Goldberg giogenic therapies, such as sorafenib, that complete or durable responses and most tu-
(sgoldber@bidmc.harvard.edu).
block signaling through multiple pathways, in- mors become refractory to treatment within
4
Division of Hematology/Oncology, Beth Israel cluding the vascular endothelial growth factor 6–12 months of initiating therapy [5, 7]. The
Deaconess Medical Center, Harvard Medical School, (VEGF) receptor [1–3]. Sorafenib is an oral mechanism of this resistance is poorly un-
Boston, MA. multikinase inhibitor of Raf-1 and is also ac- derstood, but likely includes a component of
5
tive against VEGF receptors 2 and 3, plate- non-VEGF-mediated angiogenic escape [7].
Department of Pathology, Brigham and Women’s
Hospital, Boston, MA.
let-derived growth factor receptor–β, and c- Traditionally, therapeutic response has been
KIT, giving it potent antiangiogenic ability. assessed by serial tumor size measurements,
AJR 2008; 191:133–139 It has shown significant clinical activity in most notably using the Response Evaluation
0361–803X/08/1911–133
patients with advanced renal cancer [4, 5] Criteria in Solid Tumors (RECIST) [8, 9].
and was recently approved by the U.S. Food However, preclinical assessment of new anti-
© American Roentgen Ray Society and Drug Administration for that indication tumoral therapeutics such as antiangiogenic

AJR:191, July 2008 133


agents has highlighted the limitations associ-
ated with using standard morphologic mea-
surements. Tumor response may be better
assessed by alterations in vascular perfusion
rather than tumor size, and functional mea-
surements may therefore be more appropriate
[10–12]. In addition, resistance to antiangio-
genic agents appears inevitable, but the nature
of this resistance or the ability to detect it be-
fore clinically significant disease progression
occurs remains limited. This situation high-
lights the need to develop alternative methods
to assess the effects of treatment on tumors.
Given the limitations of clinical criteria
alone, various authors have evaluated the use
of single-detector CT for the assessment of
tumor blood flow [13–25]. With the aid of reli-
able and accurate commercially available
software algorithms, they were able to show
good correlations and reproducibility of mea-
surements, thus establishing the basis for fur-
ther experimentation and the use of CT func-
tional measurements. In a more recent study
[26], perfusion MDCT achieved a better cor-
relation to the reference standard laser Dop-
pler flowmetry compared with a single-detec-
tor technique for detecting acute changes in
tumor blood flow induced over 1 hour by the
antivascular agent arsenic trioxide in an ani-
mal tumor model. In that study, perfusion
MDCT also showed potential as an indepen-
dent predictor of tumor perfusion and im-
provement in the interobserver agreement
compared with perfusion imaging using a
single-detector CT technique.
Other authors have also shown that MDCT
enables assessment of changes in tumor vas-
cularity and perfusion that result from chemo-
therapy and radiation therapy [27, 28]. Never-
theless, use of this technique over the long
time intervals required for monitoring antian-
giogenic therapy requires validation.
The purpose of this study was therefore to
determine whether perfusion MDCT can be
used to monitor the effects of antiangiogenic
therapy to potentially enable earlier prediction
of tumor response and resistance to therapy
compared with conventionally measured im-
aging parameters in an animal tumor model.
Materials and Methods
Experimental Design
The protocol was approved by the institutional
animal care and use committee (IACUC) before
study initiation. Twenty-five female rats (mean
weight, 150 ± 20 g; age range, 7–9 weeks; Fisher-344
rats, Taconic Farms) implanted with R3230
mammary adenocarcinoma (Center for Molecular
134
Sabir et al.
Imaging Research, Massachusetts General Hospi­
tal, Boston, MA) measuring 1.5 ± 0.2 cm in the
z-axis were used for this study. The experimental
group (n = 15) randomly received 7.5 or 30 mg/kg
of the antiangiogenic agent sorafenib (BAY
43-9006, Bayer) by gavage for 4 (n = 4), 9 (n = 9),
or 14 (n = 5) days. Seven untreated animals served
as the control group. This tumor model was
previously used by Hakime et al. [26] to correlate
tumor blood flow changes with perfusion MDCT
after the administration of an acute blood flow–
reducing agent.
Sorafenib Administration
Sorafenib was administered daily by gavage in
blinded fashion as 7.5 or 30 mg/kg depending on
experimental group. Sorafenib was dissolved in a
50% polyethoxylated castor oil (Cremophore EL,
Sigma)–50% ethanol mixture at four times (4×)
the desired highest concentration; for the dose of
30 mg/kg, the concentration of the dosing solution
was 8 mg/mL, with this 4× solution measuring
32 mg/mL. The compounds were heated to 60°C
for 1 minute and sonicated for 20–30 minutes to
suspend the sorafenib. Once in solution, the
aqueous component was gradually added and
diluted to generate the 1× dosing solution. The
lower dose levels were then made by dilution of
this preparation with 12.5% Cremophor EL,
12.5% ethanol, and 75% water [1]. Each dose of
sorafenib was weighed and stored in dry form
away from light and was dissolved to liquid form
immediately before administration.
Tumor Preparation
A parent tumor (≈ 1 cm in diameter) was initially
harvested from a live carrier. Within 30 minutes of
its dissection and removal, the tumor was homo­
genized with a tissue homogenizer (PowerGen
model 125, Fisher Scientific) using aseptic tech­
nique, and the tumor cells were suspended in 7 mL
of a culture medium (RPMI 1640, INC Biomedicals).
In prior control experiments in our laboratory, this
process resulted in a concentration of 1 × 107 cells
per milliliter, with more than 95% cellular viability.
During direct visualization, 0.2–0.3 mL of the
tumor cell suspension was injected slowly through an
18-gauge needle into the mammary fat pad. Animals
were monitored every 3–4 days to measure tumor
growth. Tumors were allowed to grow for 14–24 days
until the desired treatment size (1.2–2 cm) was
achieved. Tumor size was chosen to optimally match
the size with maximum z-axis coverage (4-MDCT
acquisition at 5-mm thickness = 20 mm). Tumor
induction, monitoring, and randomization were
performed by two authors. Tumors were measured
with calipers daily until reaching the desired size.
Solid tumor architecture was confirmed by sono­
graphy before CT. Tumor size and composition were
also confirmed by CT during experimentation. On
completion of the study, the animals were sacrificed
by barbiturate overdose (Somlethal, J. A. Webster)
according to IACUC guidelines.
Perfusion CT
Perfusion MDCT was performed on days 0, 4,
9, and 14 as described by Hakime et al. [26]. Rats
were scanned using a 16-MDCT scanner (Light­
Speed Plus, GE Healthcare). Before CT, the tail
vein was catheterized with a 24-gauge cannula for
administration of contrast material. After an intra­
peritoneal injection of a mixture of ketamine
(Ketaject, Phoenix Pharmaceutica), at a dose of 50
mg/kg of body weight, and xylazine (Rompun,
Bayer), at a dose of 5 mg/kg, was administered,
rats were placed on the CT table and restrained to
limit movement. When necessary, booster anes­
thetic in­jections at one tenth of those doses were ad­
min­istered intra­peritoneally every 30–60 minutes.
Initially, an unenhanced study was performed
to identify the tumor for planning purposes using
the following parameters: helical acquisition; slice
thickness, 2.5 × 2.5 mm; speed, 27 mm/s; pitch,
1.3; 120 kV; 240 mA; rotation speed, 0.5 second;
scan field of view (FOV), 24 cm; and matrix,
512 × 512. The images were then inspected by two
of the authors on the CT console; the target slices
were selected to plan subsequent dynamic studies
to ensure that the entire tumor was covered by the
20-mm scan volume.
For measurement of perfusion, CT was initiated
2 seconds before the near-instantaneous manual
administration of a 0.4-mL bolus of contrast agent
(2 mL/kg of ioversol [Optiray 350, Mallinckrodt
Imaging] at 0.05 mL/s); contrast injection was
performed consistently by the same investigator to
minimize variation. Images were obtained at
1-second intervals covering the entire tumor using
the axial mode (120 kV; 240 mA; FOV, 50 cm;
matrix, 512 × 512) throughout the bolus in­ject­ion
of contrast agent and continued for a total of 65
seconds. The 20-mm scan volume was recon­
structed into four contiguous slices collimated to
5 mm each. Immediately after the dynamic study,
the tumor was imaged in the high-resolution mode
for calculation of tumor volumes (helical
acquisition; collimation, 1.0 mm; rotation speed,
0.5 second; 120 kV; 240 mA; FOV, 24 cm; and
matrix, 512 × 512).
Image Processing
For each dynamic CT scan acquisition, four
single perfusion CT image maps, each of 5 mm
thickness, were obtained from the single scanned
20-mm tumor volume. The 400 images from the
dynamic studies were processed using perfusion
AJR:191, July 2008
 Perfusion MDCT of Antiangiogenic Therapy Response
CT software (Perfusion 2.0, GE Healthcare) with
a body tumor perfusion algorithm. This software
has been previously validated for the determination
of changes to tumor perfusion by various authors
[21–24]. A processing threshold of 0–120 H was
selected to permit appropriate subsequent analysis
of both unenhanced and enhanced soft tissue.
Arterial input was determined by placing a
region of interest (ROI) over the best visualized
artery in the slice plane (aorta, iliac artery, or
superficial femoral artery). Time–attenuation
curves were automatically generated for the arterial
input along with perfusion maps for all the tissues
within the scanning plane over the 65-second
perfusion acquisition. To determine tumor per­
fusion, an ROI was drawn freehand around the
peripheral margin of the tumor using an electronic
cursor. Care was taken to exclude peritumoral skin
and fat and intraluminal gas by viewing the cine
loop to gauge the extent of movement during
acquisition. A global time–attenuation curve for
the selected tumor tissue and the mean blood flow
for the tumor tissue within the ROI were derived.
The blood flow maps of all four single slices of
a given dynamic CT study were saved in high-
resolution gray-scale format. Every picture was
then loaded in ImageJ (Image Processing and
Analysis in Java, National Institutes of Health
Image; available at http://rsb.info.nih.gov/ij/) to
calculate the histogram of the perfusion values
within the ROIs. The pixel data from the four
tumor slices were then combined in Microsoft
Excel to calculate the mean CT perfusion for the
entire tumor volume blood flow.
Blood flow in each perfusion map was
represented in a color-coding scheme in rainbow
format such that a flow of 0 mL/min/100 g was
shown in black and maximal blood flow (50 mL/
min/100 g) was shown in bright red. Flow values
between 0 and 50 mL/min/100 g were represented
as varying shades of blue, green, yellow, and red in
order of increasing perfusion.
For calculation of tumor volumes, high-
resolution axial images were loaded as a single
volume to the workstation and were reconstructed
into multiplanar images that were used separately
by two authors to draw outlines around the tumor
edges. These outlines were used by the software
algorithm to separate the tumor from the main rat
body and to calculate its volume.
Histopathologic Correlation
All animals in the control and experiment
groups were sacrificed on day 0, 4, 9, or 14 after
their last scheduled CT scan. Tumors were sectioned
immediately (within 1 hour) in the axial plane
corresponding to the CT slice orientation, fixed in
10% formalin, and embedded in paraffin. Tissue
AJR:191, July 2008
sections were stained with H and E. All tissue
specimens were examined at their greatest axial
diameter. Nonviable tumor areas were identified
using morphologic criteria by an experienced
pathologist. Four of 18 animals in the experimental
group were exclusively followed to day 4 and were
then sacrificed for purposes of histopathologic
verification and comparison.
Statistical Analysis
Serial changes in tumor volume as determined by
CT were compared with and correlated to mean
tumor blood flow for each of the scan days 0, 4, 9,
and 14. Visual comparison of a histologic tissue
section with the corresponding animal’s perfusion
CT images was also performed. All values were
expressed as means with SDs. The Student’s t test
was used to correlate changes in tumor volume to
changes in mean tumor blood flow. A p value of 0.05
was considered to denote statistical significance.
Results
Tumor Volume
All control tumors (n = 7) increased in vol-
ume from their initial baselines values (69%
increase by day 4, 107% increase by day 9, and
143% increase by day 14) for the duration of
the experiment, whereas all treated for 9–14
days (n = 14) decreased in volume to day 9
(32% decrease from baseline) with no substan-
tial change in the rate of tumor volume shrink-
age from day 9 to day 14 (Table 1 and Fig. 1).
Accordingly, although there were no statisti-
cally significant differences in the baseline tu-
mor sizes between the control and treatment
groups (p > 0.05), there were significant differ-
ences in the rate of growth or shrinkage com-
pared with their baseline (day 0) values (p <
0.05). Similarly, significant (p < 0.05) differ-
ences in the mean tumor volumes between
animals receiving sorafenib and untreated ani-
mals at each time point were noted. No signifi-
cant differences (p > 0.1) between the high-
dose (30 mg/kg) and low-dose (7.5 mg/kg)
treatment groups in terms of the rate of change
of tumor volume with therapy were detected.
Tumor Blood Flow
Variable, but not statistically significant,
changes in mean tumor blood flow (mL/
min/100 g) were noted for the control group
through day 14 (Table 2). However, in the
sorafenib-treated groups, the mean blood flow
was significantly lower at day 4 (78% decrease
from day 0, p < 0.05) and progressed to an
average decrease of 75% for days 9 and 14
compared with day 0 (p < 0.05). However, no
statistically significant difference in the over-
all mean tumor blood flow was observed be-
tween days 4 and 9 and day 14 (p > 0.10). Fur-
thermore, there was no significant difference
(p > 0.10) be­tween the two different sorafenib
doses (7.5 and 30 mg/kg) in terms of mean
tumor blood flow and the percentage changes
from baseline.
Although there was no significant change in
either tumor volume or mean overall blood
flow when comparing sorafenib-treated day 9
or day 14 values with day 4 values, seven of 14
sorafenib-treated tumors had focal areas of new
blood flow that correlated to areas of increased
histopathologic viability despite the fact that
these tumors had shrunk or were stable in size
(day 4, 2.18 ± 0.8 cm3; day 9, 1.98 ± 0.8 cm3).
These foci were seen in four of nine (44%) tu-
mors from day 4 to day 9 and in three of five
(60%) tumors from day 9 to day 14 (Fig. 2).
Thus, 50% of the sorafenib-treated tumors ex-
hibited new foci of increased perfusion from
day 4 onward as seen from spatial analysis of
the individual perfusion CT slices. These new
foci of increased perfusion were seen on the
middle two slices of the four slices covering
each tumor in five of seven (71%) tumors and
on one of four and three of four slices in the
other two tumors (2/7, 29%).
Histopathologic Correlation
Microscopic analysis of tissue sections
revealed a close correlation between the per-
fusion CT blood flow maps and histopatho-
logic features (Fig. 3). We observed small
areas of central necrosis within the control
tumors and larger areas of cell death in every
tumor treated with 4, 9, or 14 days of
sorafenib. Based on this analysis, we were
able to correlate the color coding of the blood
flow maps to the degree of viability of tumor
cells. Specifically, a threshold of 2.5 mL/
min/100 g represented by light blue on the
color map indicated the presence of viable
cells as seen by microscopic comparison.
This threshold and color coding correlated
well for representing cell viability versus cell
death across both the control group and the
sorafenib-treated tumors. Moreover, a thres­
hold of 0–1 mL/min/100 g was associated
with the presence of nonviable areas within
the tumor across both the control group and
sorafenib-treated tumors.
Comparison of Tumor Volume and Mean
Tumor Blood Flow
Poor correlation was seen between chang-
es in blood flow and tumor volume for days
0–9 (r 2 = 0.34), 4–9 (r 2 = 0.0004), and 9–14
135
 Sabir et al.

TABLE 1: Changes in Volume for Control and Sorafenib-Treated R3230 Tumors

Tumor Volume
Day 4 Day 9 Day 14
No. of Day 0 % % %
Group Rats (Mean ± SD) (cm3) Mean ± SD (cm3) Changea Mean ± SD (cm3) Changea Mean ± SD (cm3) Changea
Control 7 2.0 ± 0.7 3.5 ± 1.5 69 4.3 ± 1.8 107 5.9 ± 1.0 143
Sorafenib (all rats) 14 2.5 ± 1.1 1.9 ± 1.0 –24 1.7 ± 0.8 –32 2.1 ± 1.0 –31
Sorafenib (all rats, days 0–9) 9 2.3 ± 1.2 1.7 ± 1.0 –23 1.5 ± 0.80 –32 — —
High dose 4 1.7 ± 0.8 1.7 ± 1.0 –5 1.4 ± 0.6 –14 — —
Low dose (days 0–9) 5 2.7 ± 1.4 1.8 ± 1.2 –37 1.5 ± 1.0 –47 — —
Low dose (days 0–14) 5 2.9 ± 0.8 2.2 ± 0.8 –26 2.0 ± 0.8 –33 2.1 ± 1.0 –31
Note—Percentage change is reported for the mean as compared to baseline (day 0). As a rule, control tumors consistently increased in volume whereas the sorafenib-
treated tumors decreased in volume. Dash (—) indicates not applicable.
aPercentage change values represent the mean of changes calculated on an animal-by-animal basis, not the percent change of the means.

(r 2 = 0.16). Thus, we were unable to establish Discussion
a clear-cut pattern of increase or decrease in We evaluated the use of perfusion MDCT
blood flow, tumor volume, or both correlat- as an alternative means of determining the
ing with progression at 14 days of therapy. blood flow changes associated with antiangio-
A B
genic therapy in a well-established and well-
studied animal tumor model. CT has emerged
recently as a means of rapid, noninvasive as-
sessment not only of anatomy, but also of
physiology, as is evident in its ability to reli-
ably measure blood flow [13, 14, 26]. MDCT
is fast, economical, and widely available, all
of which contribute to its enormous potential
in revolutionizing the care of patients with
cancer, particularly those who receive antian-
giogenic agents such as sorafenib.
Hakime et al. [26] previously showed the
superiority of MDCT to single-detector CT
for measuring tumor blood flow, as evidenced
by its higher interobserver agreement and
greater tumor volume coverage. Our study
results further point to the fact that the cur-
rently available systems of tumor monitor-
ing, such as RECIST, alone cannot be en-
tirely relied on when using agents that reduce
tumor blood flow because changes in blood
flow may precede changes in tumor size. Our
findings also reveal that a set of tumors re-
ceiving the antiangiogenic agent sorafenib
can exhibit differences in response to that
agent over time, which can range from
changes in mean tumor blood flow to devel-
opment of new zones of perfusion in areas

Fig. 1—Perfusion MDCT maps of control group

tumors (outlined regions and arrows) show
progressive increase in size, volume, and blood flow
from days 0–14.
A–D, Images obtained at days 0 (A), 4, (B), 9 (C),
and 14 (D). Blood flow in each perfusion map is
represented in color-coding scheme in rainbow
format such that flow of 0 mL/min/100 g is shown in
black and maximal blood flow (50 mL/min/100 g) is
shown in bright red. Flow values between 0 and 50
mL/min/100 g are represented as varying shades of
blue, green, yellow, and red in order of increasing
perfusion.
C D

136 AJR:191, July 2008

 Perfusion MDCT of Antiangiogenic Therapy Response

that initially showed minimal or no blood
flow. We found that these foci of new perfu-
sion can develop and be detected by perfu-
sion CT even when mean tumor blood flow is
unchanged or while tumor size or volume is
actively decreasing, greatly emphasizing the
importance of perfusion MDCT in such a
scenario. Such zones of blood flow corre-
spond to focal areas of viable tumor and may
well represent the development of new ves-
sels that occur despite the administration of
the antiangiogenic agent, indicating the
emergence of cells that have developed a
means of resistance to the therapeutic agent.
Furthermore, the fact that the zones of new
perfusion were not seen in every image fur-
ther emphasizes the importance of covering
large volumes of the tumor with MDCT
techniques to maximize diagnostic sensitiv-
ity. An alternative approach could be to scan
and calculate the blood flow in the middle
two slices (center part) of the tumor only in
cases in which full tumor volume coverage is
not possible. In addition, perfusion CT will

TABLE 2: Mean Tumor Blood Flow Values for Control and Sorafenib-Treated R3230 Tumors
Blood Flow
Day 4 Day 9 Day 14
Day 0
No. of (Mean ± SD) Mean ± SD % Mean ± SD % Mean ± SD %
Group Rats (mL/min/100 g) (mL/min/100 g) Changea (mL/min/100 g) Changea (mL/min/100 g) Changea
Control 7 18.1 ± 9.2 15.8 ± 5.6 –13 21.7 ± 12.2 20 27.7 ± 34 78
Sorafenib (all rats) 14 23.8 ± 11.6 5.2 ± 3.2 –78 6.4 ± 4.0 –66 6.3 ± 5.2 –83
Sorafenib (all rats, days 0–9) 9 16.8 ± 6.6 4.0 ± 2.4 –75 6.3 ± 4.6 –60 — —
High dose 4 17.6 ± 6.0 3.1 ± 2.3 –82 4.4 ± 2.0 –75 — —
Low dose (days 0–9) 5 16.2 ± 7.7 4.7 ± 2.5 –70 7.9 ± 5.7 –48 — —
Low dose (days 0–14) 5 34.3 ± 9.4 7.3 ± 3.4 –80 5.4 ± 4.0 –78 5.3 ± 5.3 –83
Note—Percentage change is reported for the mean as compared to baseline (day 0). As a rule, control tumors maintained their blood flow whereas the sorafenib-treated
tumors showed marked decrease in mean tumor blood flow that remained stable from day 9 onward into therapy. Dash (—) indicates not applicable.
aPercentage change values represent the mean of changes calculated on an animal-by-animal basis, not the percent change of the means.

A B C

D E F
Fig. 2—Perfusion CT maps of two rats from sorafenib-treated group show decrease in tumor volume and blood flow from days 0–9.
A–F, Maps of rat 1 (A–C) and rat 2 (D–F) at days 0 (A and D), 4 (B and E), and 9 (C and F). Blood flow in each perfusion map is represented in color-coding scheme in
rainbow format such that flow of 0 mL/min/100 g is shown in black and maximal blood flow (50 mL/min/100 g) is shown in bright red. Flow values between 0 and 50 mL/
min/100 g are represented as varying shades of blue, green, yellow, and red in order of increasing perfusion. In addition, these figures show development of rim of
increased perfusion (arrows, C and F) in sorafenib-treated tumors despite continued decrease in tumor volume. This could possibly represent early development of tumor
resistance to antiangiogenic therapy.

AJR:191, July 2008 137


Fig. 3—Histopathologic correlation (right) of sorafenib-treated tumor with CT map (left). Areas of increased
and decreased or absent blood flow (yellow arrows) on perfusion images correspond to viable and nonviable
tumor areas (black arrows), respectively, on tissue section.
ultimately need to be compared with other
imaging strategies for mapping perfusion
such as arterial spin-labeling MRI [29] and
dynamic contrast-enhanced Doppler sonog-
raphy [30].
Perfusion MDCT was able to successfully
identify focal areas of new tumor perfusion
9–14 days into antiangiogenic therapy before
clinically measurable changes in tumor mass
were seen, perhaps indicating early angio-
genic escape in this specific animal tumor
model. However, not all tumors may behave
in an equal manner with antiangiogenic ther-
apy, and thus further experimentation in
other animal, and more importantly human,
tumor models is needed to validate the ef-
fectiveness of perfusion MDCT for this spe-
cific purpose. Thus, specific clinical testing
in RCC, a disease that is known to be sensi-
tive to sorafenib, would be of interest and
may or may not reproduce these findings.
Although we were able to correlate the mi-
croscopy findings with the perfusion CT
maps to within reasonable limits, our study is
limited by the resolution and slice thickness
used for calculating tumor blood flow. This
can be overcome in future studies by the
availability of software that allows higher-
resolution scanning and use of thinner slices,
thus enabling tighter correlation with histo-
pathology results. As an additional limitation,
we relied on conventional histologic criteria
of cellular viability rather than microvascular
density. Nevertheless, tumor growth and via-
bility, not angiogenesis, are the primary end
138
Sabir et al.
points or concerns. Furthermore, Kan et al.
[13] have shown that “functional CT can help
quantify the perfusion function of mature
vessels but not changes in microvessel density
in antiangiogenic therapy.” Perfusion CT of
thoracic and abdominal tumors may be lim-
ited by breathing motion; nonetheless, previ-
ous studies [26] using similar techniques and
an animal model have shown the accuracy of
perfusion CT compared with an established
reference standard. Future advances in imag-
ing software and technology will help to fur-
ther optimize image acquisition and process-
ing with lower radiation doses.
Several potential limitations in our animal
model and experimental design warrant dis-
cussion. Like most other scientific measure-
ment systems, perfusion CT has room for
variation, particularly for heterogeneous bio-
logic systems such as orthotopic tumors. For
example, we note that there was a 13% drop
in blood flow in the control rats on day 4
compared with their baseline values, but this
decrease is well below the reported coeffi-
cient of variability in perfusion CT measure-
ments [18, 20, 29]. Furthermore, we did not
sacrifice control rats on days 0 and 4 for this
study. However, prior experiments in our
laboratory [26] using the exact same animal
tumor model have shown that untreated
R3230 tumors of similar sizes have excellent
solidity and vascularity as indicated by ra-
diologic–pathologic correlation. Moreover,
our end point was to determine the vascular-
ity of the tumors at days 9 and 14, not at day
4. This, combined with the limited number
of rats available, led to an experimental de-
sign that did not sacrifice control animals on
days 0 and 4.
Inevitably, although rodent studies allow
minimization of variability of tumor parame-
ters, the size of the model may introduce un-
toward errors in the imaging technique. For
example, perfusion CT is likely to work best
with a tight injection bolus. Nevertheless, by
necessity, we used a near-instantaneous man-
ual bolus of 0.4 mL of contrast material, as
done in previous published studies [26]. Even
though a constant injection rate using a power
injector is perhaps the ideal method, to the
best of our knowledge, there is no power in-
jector commercially available for use in the
tail vein of a 150-g rat. Furthermore, although
rats were positioned almost identically every
time they were scanned, this position change
might have led to variability in tumor volume
measurements for such small tumors. How-
ever, unpublished data from our laboratory
show that repeated volume CT scans of the
same rat tumor with the rat in varying posi-
tions (prone, supine, lateral, and prone again)
have average variability in calculated tumor
volumes of 1.0% ± 0.5% (range, –1.9% to
1.2%). Accordingly, this degree of error is
way below the changes in tumor volume ob-
served in this study.
Lastly, some authors have used additional
perfusion CT calculation parameters such as
blood volume, mean transit time, and perme-
ability surface area for determination of a
tumor’s vascularity in varying tumor models
[31, 32]. However, studies from our labora-
tory [26] have clearly shown tight correlation
between blood flow as measured by perfu-
sion MDCT and laser Doppler flowmetry as
the reference standard in the same animal
model. Thus, we preferentially used this pa-
rameter over those with which poorer corre-
lations (unpublished data) were found. More-
over, various authors have clearly shown that
of all the commonly calculated perfusion CT
parameters (i.e., blood flow, blood volume,
mean transit time, and permeability surface
area product), blood flow shows the closest
correlation to histopathology findings and
clinical outcomes [13, 31, 32].
If successfully validated, perfusion MDCT
could be used to further characterize the
phenomenon of “early breakthrough” in
large-scale trials in patients with highly vas-
cular tumors such as RCC, thus enabling us
to gain more insight into the physiology and
pathogenesis of resistant tumor cells. Earlier
AJR:191, July 2008
 Perfusion MDCT of Antiangiogenic Therapy Response
identification of resistance through measure-
ment of blood flow would enable clinicians
to consider an alteration in therapy, including
dose or schedule modifications or switching
to another agent before tumor growth is doc-
umented, possibly preventing or delaying
clinical sequelae of disease progression.
Moreover, tumors exhibiting this phenome-
non could then be easily singled out for tar-
geted biopsy and extraction of the resistant
cells, which can be biochemically and ge-
netically studied to understand and identify
the exact resistance mechanism or mecha-
nisms, thus enabling future development of
targeted therapies.
In conclusion, in this murine xenograft tu-
mor model receiving antiangiogenic therapy,
perfusion MDCT was able to identify chang-
es in tumor blood flow before changes in tu-
mor volume occurred, possibly indicating
early reversal of tumor responsiveness to
therapy. Given that changes in tumor volume
after antiangiogenic therapy do not necessar-
ily correlate with true treatment response,
noninvasive physiologic imaging of tumor
perfusion may be necessary for evaluating
the effectiveness of antiangiogenic agents.
References
1. Wilhelm SM, Carter C, Tang L, et al. BAY
43-9006 exhibits broad spectrum oral antitumor
activity and targets the RAF/MEK/ERK pathway
and receptor tyrosine kinases involved in tumor
progression and angiogenesis. Cancer Res 2004;
64:7099–7109
2. Flaherty KT. Sorafenib in renal cell carcinoma.
Clin Cancer Res 2007; 13(2 Pt 2):747S–752S
3. Hutson TE, Sonpavde G, Galsky MD. Targeting
growth factor and antiangiogenic pathways in
clear-cell renal cell carcinoma: rationale and on-
going trials. Clin Genitourin Cancer 2006;
5[suppl 1]:S31–S39
4. Ratain MJ, Eisen T, Stadler WM, et al. Phase II
placebo-controlled randomized discontinuation
trial of sorafenib in patients with metastatic renal
cell carcinoma. J Clin Oncol 2006; 24: 2505–
2512
5. Escudier B, Eisen T, Stadler WM, et al.; TAR-
GET Study Group. Sorafenib in advanced clear-
cell renal-cell carcinoma. N Engl J Med 2007;
356:125–­134 [erratum in N Engl J Med 2007;
357:203]
6. FDA approves new treatment for advanced kidney
cancer. FDA News. P05-107. Published December
20, 2005. http://www.fda.gov/bbs/topics/news/
AJR:191, July 2008
2005/new01282.html. Accessed March 9, 2007
7. Sosman JA, Puzanov I, Atkins MB. Opportunities
and obstacles to combination targeted therapy in
renal cell cancer. Clin Cancer Res 2007; 13(2 Pt
2):764S–769S
8. World Health Organization. WHO handbook for
reporting results of cancer treatment. Geneva,
Switzerland: World Health Organization, 1979:
WHO offset publication no. 48
9. Therasse P, Arbuck SG, Eisenhauer EA, et al.
New guidelines to evaluate the response to treat-
ment in solid tumors. European Organization for
Research and Treatment of Cancer, National Can-
cer Institute of the United States, National Cancer
Institute of Canada. J Natl Cancer Inst 2000;
92:205–216
10. Li WW. Tumor angiogenesis: molecular patholo-
gy, therapeutic targeting and imaging. Acad Ra-
diol 2000; 7:800–811
11. Grosios K, Holwell SE, McGown AT, Pettit GR,
Bibby MC. In vivo and in vitro evaluation of com-
bretastatin A-4 and its sodium phosphate prodrug.
Br J Cancer 1999; 81:1318–1327
12. Chaplin DJ, Pettit GR, Hill SA. Anti-vascular ap-
proaches to solid tumor therapy: evaluation of
combretastatin A4 phosphate. Anticancer Res
1999; 19:189–195
13. Kan Z, Phongkitkarun S, Kobayashi S, et al.
Functional CT for quantifying tumor perfusion in
antiangiogenic therapy in a rat model. Radiology
2005; 237:151–158
14. Lee TY, Purdie TG, Stewart E. CT imaging of an-
giogenesis. Q J Nucl Med 2003; 47:171–187
15. Kan Z, Kobayashi S, Phongkitkarun S, Charnsan-
gavej C. Functional CT quantification of tumor
perfusion after transhepatic arterial embolization
in a rat model. Radiology 2005; 237:144–150
16. Fournier LS, Cuenod CA, de Bazelaire C, et al.
Early modifications of hepatic perfusion mea-
sured by functional CT in a rat model of hepato-
cellular carcinoma using a blood pool contrast
agent. Eur Radiol 2004; 14:2125–2133
17. Willett CG, Boucher Y, Di Tomaso E, et al. Direct
evidence that the VEGF-specific antibody bevaci-
zumab has antivascular effects in human rectal
cancer. Nat Med 2004; 10:145–147
18. Nabavi DG, Cenic A, Dool J, et al. Quantitative
assessment of cerebral hemodynamics using CT:
stability, accuracy, and precision studies in dogs. J
Comput Assist Tomogr 1999; 23:506–515
19. Nabavi DG, Cenic A, Craen RA, et al. CT assess-
ment of cerebral perfusion: experimental valida-
tion and initial clinical experience. Radiology
1999; 213:141–149
20. Cenic A, Nabavi DG, Craen RA, Gelb AW, Lee
TY. A CT method to measure hemodynamics in
brain tumors: validation and application of cere-
bral blood flow maps. Am J Neuroradiol 2000;
21:462–470
21. Purdie TG, Henderson E, Lee TY. Functional CT
imaging of angiogenesis in rabbit VX2 soft-tissue
tumor. Phys Med Biol 2001; 46:3161–3175
22. Gillard JH, Antoun NM, Burnet NG, Pickard JD.
Reproducibility of quantitative CT perfusion im-
aging. Br J Radiol 2001; 74:552–555
23. Tsushima Y, Funabasama S, Aoki J, Sanada S,
Endo K. Quantitative perfusion map of malignant
liver tumors, created from dynamic computed to-
mography data. Acad Radiol 2004; 11:215–223
24. Tsushima Y, Unno Y, Koizumi J, Kusano S. He-
patic perfusion changes after transcatheter arteri-
al embolization (TAE) of hepatocellular carcino-
ma: measurement by dynamic computed
tomography (CT). Dig Dis Sci 1998; 43:317–322
25. Lin WY, Wang SJ, Yeh SH. Hepatic perfusion in-
dex in evaluating treatment effect of transcatheter
hepatic artery embolization in patients with hepa-
tocellular carcinoma. Neoplasma 1995; 42:89–92
26. Hakime A, Peddi H, Hines-Peralta AU, et al. CT
perfusion for determination of pharmacologically
mediated blood flow changes in an animal tumor
model. Radiology 2007; 243:712–719
27. Sahani DV, Kalva SP, Hamberg LM, et al. Assess-
ing tumor perfusion and treatment response in
rectal cancer with multisection CT: initial obser-
vations. Radiology 2005; 234:785–792
28. Meijerink MR, van Cruijsen H, Hoekman K, et al.
The use of perfusion CT for the evaluation of
therapy combining AZD2171 with gefitinib in
cancer patients. Eur Radiol 2007; 17:1700–1713
29. De Bazelaire C, Rofsky NM, Duhamel G, Mi-
chaelson MD, George D, Alsop DC. Arterial spin
labeling blood flow magnetic resonance imaging
for the characterization of metastatic renal cell
carcinoma (1). Acad Radiol 2005; 12:347–357
30. Lamuraglia M, Escudier B, Chami L, et al. To
predict progression-free survival and overall sur-
vival in metastatic renal cancer treated with
sorafenib: pilot study using dynamic contrast-
enhanced Doppler ultrasound. Eur J Cancer
2006; 42:2472–2479 [Erratum in Eur J Cancer
2007; 43:1336]
31. Sahani DV, Holalkere NS, Mueller PR, Zhu AX.
Advanced hepatocellular carcinoma: CT perfu-
sion of liver and tumor tissue—initial experience.
Radiology 2007; 243:736–743
32. Hermans R, Lambin P, Van der Goten A, et al.
Tumoural perfusion as measured by dynamic
computed tomography in head and neck carcino-
ma. Radiother Oncol 1999; 53:105–111
139