Urinary Bladder Cancer:

Radiology

Willem M. L. L. G. Deserno,
MD, MSc
Mukesh G. Harisinghani, MD Preoperative Nodal Staging
Matthias Taupitz, MD, PhD
Gerrit J. Jager, MD, PhD
J. Alfred Witjes, MD, PhD
with Ferumoxtran-10 –
Peter F. Mulders, MD, PhD
Christina A. Hulsbergen van
enhanced MR Imaging1
de Kaa, MD, PhD
D. Kaufmann, MD PURPOSE: To prospectively evaluate ferumoxtran-10 – enhanced magnetic reso-
Jelle O. Barentsz, MD, PhD nance (MR) imaging for nodal staging in patients with urinary bladder cancer.
Index terms:
Bladder neoplasms, metastases, MATERIALS AND METHODS: Fifty-eight patients with proved bladder cancer
83.31, 83.33 were enrolled. Results of MR imaging performed before and after injection of
Bladder neoplasms, MR, 83.12141 ferumoxtran-10 were compared with histopathologic results in surgically removed
Iron lymph nodes. High-spatial-resolution three-dimensional T1-weighted magnetiza-
Magnetic resonance (MR), contrast
media
tion-prepared rapid acquisition gradient-echo (voxel size, 1.4 ⫻ 1.4 ⫻ 1.4 mm) and
T2*-weighted gradient-echo (voxel size, 0.8 ⫻ 0.8 ⫻ 3.0 mm) sequences were
Published online before print performed before and 24 hours after injection of ferumoxtran-10 (2.6 mg iron per
10.1148/radiol.2332031111 kilogram of body weight). On precontrast images, lymph nodes were defined as
Radiology 2004; 233:449 – 456
malignant by using size and shape criteria (round node, ⬎8 mm; oval, ⬎10 mm
axial diameter). On postcontrast images, nodes were considered benign if there was
1
From the Departments of Radiol- homogeneous decrease in signal intensity and malignant if decrease was absent or
ogy (W.M.L.L.G.D., G.J.J., J.O.B.),
Urology (J.A.W., P.F.M.), and Pathol- heterogeneous. Qualitative evaluation was performed on a node-to-node basis.
ogy (C.A.H.v.d.K.), University Medical Sensitivity, specificity, predictive values, and accuracy were evaluated with logistic
Center Sint Radboud, PO Box 9101, 6500 regression analysis.
HB Nijmegen, the Netherlands; Depart-
ments of Radiology (M.G.H.) and On-
cology (D.K.), Massachusetts General RESULTS: In 58 patients, 172 nodes imaged with use of ferumoxtran-10 were
Hospital, Boston, Mass; and Depart- matched and correlated with results of node dissection. Of these, 122 were benign
ment of Radiology, Charité Hospital Ber- and 50 were malignant. With nodal size and shape criteria, accuracy, sensitivity,
lin, Germany (M.T.). Received July 15,
2003; revision requested September 18; specificity, and positive and negative predictive values on precontrast images were
final revision received February 24, 92%, 76%, 99%, 97%, and 91%, respectively; corresponding values on postcon-
2004; accepted March 16. Address
correspondence to W.M.L.L.G.D. (e- trast images were 95%, 96%, 95%, 89%, and 98%. In the depiction of pelvic
mail: w.deserno@rad.umcn.nl). metastases, sensitivity and negative predictive value improved significantly at post-
Authors stated no financial relation- contrast compared with those at precontrast imaging, from 76% to 96% (P ⬍ .001)
ship to disclose. and from 91% to 98% (P ⬍ .01), respectively. At postcontrast imaging, metastases
Author contributions: (4 –9 mm) were prospectively found in 10 of 12 normal-sized nodes (⬍10 mm);
Guarantors of integrity of entire these metastases were not detected on precontrast images. Postcontrast images also
study, W.M.L.L.G.D., M.G.H., M.T.,
G.J.J., J.A.W., P.F.M., C.A.H.v.d.K., showed lymph nodes that were missed at pelvic node dissection in two patients.
J.O.B.; study concepts, W.M.L.L.G.D.,
M.G.H., M.T., J.O.B.; study design, CONCLUSION: Ferumoxtran-10 – enhanced MR imaging significantly improves
W.M.L.L.G.D., M.G.H., J.O.B.; litera- nodal staging in patients with bladder cancer by depicting metastases even in
ture research, W.M.L.L.G.D., M.G.H.,
G.J.J., J.A.W., J.O.B.; clinical studies, normal-sized lymph nodes.
W.M.L.L.G.D., M.G.H., M.T., J.A.W., © RSNA, 2004
P.F.M., C.A.H.v.d.K., D.K., J.O.B.; data ac-
quisition, W.M.L.L.G.D., M.G.H., M.T.,
J.A.W., P.F.M., C.A.H.v.d.K., D.K., J.O.B.;
data analysis/interpretation, all authors;
statistical analysis, W.M.L.L.G.D., G.J.J.;
manuscript preparation, W.M.L.L.G.D.,
J.O.B.; manuscript definition of intellec-
tual content, W.M.L.L.G.D., M.G.H., Cancer of the urinary bladder is one of the most common types of malignant tumor of the
M.T., G.J.J., J.A.W., P.F.M., C.A.H.v.d.K., urinary tract. Because the sequence of surgical or systemic treatment and the prognosis
J.O.B.; manuscript editing, M.G.H., G.J.J.,
J.A.W., C.A.H.v.d.K., J.O.B.; manuscript re- depend on the depth of tumor infiltration and the extent of metastatic lymph nodes (1),
vision/review, M.G.H., M.T., G.J.J., J.A.W., it is important to accurately assess the stage of the nodes prior to surgery. With current
P.F.M., C.A.H.v.d.K., J.O.B.; manuscript cross-sectional imaging modalities such as computed tomography (CT) and magnetic
final version approval, W.M.L.L.G.D.,
M.G.H., J.O.B. resonance (MR) imaging, we rely predominantly on nodal size for detecting metastases.
© RSNA, 2004 However, there is considerable overlap in size between benign and malignant nodes. By
using nodal size and morphology criteria, Jager et al (2) found that oval nodes with a

449
 diameter larger than 10 mm and round
nodes with a diameter larger than 8 mm
could be characterized as malignant with
a sensitivity of 83% and specificity of
98%. With these criteria, however, the au-
Radiology

thors were unable to detect lymph node

metastasis in oval nodes smaller than 10
mm or in round nodes smaller than 8 mm
in diameter. Gadolinium-enhanced MR
imaging is not useful for detecting nodal
metastases because normal-sized meta-
static nodes and nonmetastatic nodes may
show similar enhancement (3). The role of
whole-body positron emission tomogra-
phy (PET) with fluorine 18 fluorodeoxyglu-
cose, or FDG, in nodal staging is limited
because FDG accumulates as part of the
physiologic process in the urinary bladder
and thus obscures PET-positive nodes; also,
the level of FDG uptake in bladder cancer is
low (4).
Thus, to detect metastasis in normal-
sized lymph nodes it is necessary to ob-
tain differential enhancement between
normal lymph node tissue and metasta-
ses. Results of previous studies in which
ultrasmall superparamagnetic iron oxide
(ferumoxtran-10) particles were used
have shown that normal nodal tissue
shows contrast material uptake and a se-
lective decrease in signal intensity on T2-
or T2*-weighed MR images, whereas
nodal areas infiltrated with metastases
lack ferumoxtran-10 uptake and retain
high signal intensity on ferumoxtran 10 –
enhanced MR images (Fig 1) (5,6). How-
ever, authors of these studies did not
comment on the role of ferumoxtran-10
in the detection of metastases in nodes
smaller than 10 mm (6). In a study of Figure 1. Drawing shows uptake of ferumoxtran-10. Injected particles slowly extravasate from
prostate cancer, it was shown that feru- the vascular to the interstitial space (1) and then are transported to lymph nodes via lymphatic
moxtran 10 – enhanced MR imaging al- vessels (2). In lymph nodes, particles are internalized by macrophages (3), and these intracellular
lowed the detection of small (⬍10-mm) iron-containing particles cause changes in magnetic properties detectable at MR imaging. The
iron-loaded macrophages cause normal nodal tissue to have low signal intensity on MR images.
metastatic lymph nodes (7), which re-
Disturbances of lymph flow or nodal architecture by metastases lead to abnormal accumulation
sulted in a significant increase in sensi- patterns depicted at MR imaging by the lack of decreased signal intensity (4).
tivity. Metastatic nodes in bladder carci-
noma, however, are larger (2); therefore,
the results found for prostate cancer can-
not be extrapolated to urinary bladder (n ⫽ 8), were enrolled in this prospective MR Imaging
cancer. The purpose of our study was to study (Table 1). Patients were excluded if
MR imaging was performed at 1.5 T
prospectively evaluate ferumoxtran 10 – they had hemochromatosis or an allergy
(Magnetom Vision, Siemens, Erlangen,
enhanced MR imaging for nodal staging to iron compounds or if they were preg-
in patients with urinary bladder cancer. nant or breast-feeding. Prior to MR imag- Germany; or Horizon, GE Medical Sys-
ing, all patients underwent cystoscopy tems, Milwaukee, Wis) with a pelvic
and subsequent transurethral biopsy. The phased-array coil. Imaging was per-
MATERIALS AND METHODS mean interval between biopsy and MR formed before and 24 –36 hours after in-
imaging was 18 days (range, 10 –28 days). travenous infusion of ferumoxtran-10
Patients (particle size, 35 nm; Sinerem, Guerbet,
Local staging was based on clinical exam-
Fifty-eight patients with proved uri- ination and biopsy results according to Paris, France; or Combidex, Advanced
nary bladder cancer who were consecu- the TNM staging system of the American Magnetics, Cambridge, Mass), which was
tively scheduled to undergo cystectomy Joint Committee on Cancer (8). The administered at a dose of 2.6 mg of iron
at the University Medical Center Nijme- study was approved by the medical ethics (Fe) per kilogram of body weight. MR
gen (n ⫽ 32), Massachusetts General Hos- committees at each institution, and all images were acquired by using high-
pital (n ⫽ 18), or Charité Hospital Berlin patients gave written informed consent. spatial-resolution three-dimensional T1-

450 䡠 Radiology 䡠 November 2004 Deserno et al


TABLE 1
Patient Characteristics
Characteristic
No. of patients
Radiology
Sex
Male
Female
Median age (y)
Procedure performed
Open lymph node dissection
Laparoscopic lymph node dissection
Image-guided biopsy
Pathologic stage†
T1
T2
T3
T4
* Data in parentheses are percentages.
† Data are percentages of patients.
Figure 2. Photograph shows dissected nodes
placed on a grid to indicate their former loca-
tion in the patient.
weighted magnetization-prepared rapid
acquisition gradient-echo (repetition time
msec/echo time msec, 11.4/4.4; flip angle,
8°; section thickness, 1.4 mm) and two-
dimensional T2*-weighted fast low-angle
shot or T2*-weighted multiple echo data
imaging combination (MEDIC; Siemens)
sequences (800–1500/25.4; flip angle, 30°;
section thickness, 3.0 mm). For all se-
quences, images were obtained in the
transverse plane; for the T2*-weighted se-
quences, images were additionally ac-
quired in planes parallel to the iliac vessels
(obturator planes). Side effects of ferumox-
tran-10 were recorded.
Nodal Assessment
A node-to-node analysis was per-
formed by using the pre- and postcon-
trast images. The MR images were pro-
Volume 233 䡠 Number 2
Value*
58
48
10
60
44 (76)
2 (3)
12 (21)
4 tocols, the urologist removed all lymph
20
57
19
spectively analyzed by one experienced
observer at each of the participating in-
stitutions (J.O.B., M.T., or M.G.H., with
16, 13, or 5 years of experience, respec-
tively); observers were unaware of the
clinical stage or histologic grade. On pre-
contrast MR images, lymph nodes were
rated as malignant on the basis of nodal
size and shape criteria as described by
Jager et al (2). The distinction between
round and oval nodes was made on the
basis of the ratio between the short axis
and long axis. A node was called round if
this ratio was 0.8 –1.0. A node was called
oval if this ratio was smaller than 0.8. The
three-dimensional sequences were evalu-
ated on soft-copy images by using multi-
planar reconstruction at a workstation.
An oval node was considered metastatic
if the minimal axial diameter was greater
than 10 mm, and a round node was con-
sidered metastatic if the minimal axial
diameter was greater than 8 mm. On the
ferumoxtran-10 – enhanced MR images, a
node was considered normal if it showed
a homogeneous signal intensity decrease
and was considered metastatic if the en-
tire node or a focal area did not show a
signal intensity decrease on T2*-weighted
images. First, the precontrast images were
read with use only of the objective nodal
size and shape criteria described by Jager et
al. Thereafter, the combined pre- and post-
contrast MR images were read, and judg-
ment was based only on changes in signal
intensity.
Forty-four patients (76%) underwent
pelvic lymph node dissection, 12 pa-
tients (21%) underwent image-guided bi-
opsy, and two patients (3%) underwent
laparoscopic lymph node dissection. Im-
age-guided biopsy was performed only in
nodes larger than 8 mm. To ensure
Preoperative Nodal Staging of Urinary Bladder Cancer
“node-to-node” comparison, the urolo-
gists (J.A.W., P.F.M.) were preoperatively
provided with a schematic drawing of all
visible nodes on MR images in relation-
ship to fixed anatomic landmarks, such
as the iliac arteries or ureters. To further
ensure correct nodal correlation, the ra-
diologist (W.M.L.L.G.D., J.O.B., M.G.H.,
or M.T.) was present at the time of sur-
gery and helped the surgeon to identify
the nodes with the use of marked grids.
In accordance with standard surgical pro-
nodes in the obturator and the common
iliac regions on both sides. All additional
nodes, which were removed on the basis
of findings on the ferumoxtran-10 – en-
hanced MR images, were noted sepa-
rately.
After removal, the dissected nodes
were placed on a grid to show their
former location (Fig 2) and thereafter
were sent for pathologic analysis. The pa-
thologist (C.A.H.v.d.K., 11 years of expe-
rience in bladder cancer) described nodal
size in two dimensions and indicated
both normal and metastatic nodes on a
schematic drawing that was identical to
the one provided to the urologist prior to
the operation. The histopathologic ex-
amination of the nodes included hema-
toxylin-eosin staining. Finally, the radi-
ologist and pathologist performed the
node-to-node correlation in consensus
by comparing the node location and di-
agnosis determined at MR imaging with
the pathologist’s report. All results were
independently evaluated by several au-
thors (W.M.L.L.G.D., J.O.B., M.G.H., and
M.T.).
Statistical Analysis
Power analysis was performed by as-
suming a precontrast imaging sensitivity
of about 60% and a postcontrast imaging
sensitivity of 90%. The number of match-
ing patients becomes 25 for each group.
If we assume a 10% drop-out rate, we
needed a minimum of 58 patients for a P
value of .05 and a power of .80. Sensitiv-
ity, specificity, and negative and positive
predictive values for pre- and postcon-
trast MR images were determined. P ⬍ .01
was considered to indicate a statistically
significant difference. All results were an-
alyzed by using logistic regression analy-
sis. The ␹2 test was used only for descrip-
tive statistics. Statistical analysis was
carried out by using a software package
(SPSS version 10.1; SPSS, Chicago, Ill).
䡠 451
 RESULTS
TABLE 2
Results at Precontrast MR Imaging and Node-to-Node Comparison with
In two patients, minor side effects were Histologic Findings
seen after the start of ferumoxtran-10 in-
fusion, in the form of minor back pain Histologic Findings
Precontrast Imaging
Radiology

that disappeared after the infusion was Results Positive Negative All Nodes
stopped. After the infusion was reiniti- Positive 38 1 39
ated a few minutes later, the symptoms Negative 12 121 133
did not reappear. No medical treatment Total 50 122 172
was needed.
At pelvic lymph node dissection, an Note.—Positive results indicate metastatic nodes, and negative results indicate nonmetastatic
nodes. For precontrast MR imaging, accuracy was 92%; sensitivity, 76%; specificity, 99%; negative
average of nine nodes per patient (range, predictive value, 91%; and positive predictive value, 97%.
two to 21 nodes per patient) were re-
moved.
In 172 (43%) of 404 dissected nodes,
an accurate match could be made be- TABLE 3
tween histologic evaluation and MR im- Results at Postcontrast MR Imaging and Node-to-Node Comparison with
ages (Tables 2, 3). The remaining 232 Histologic Findings
nodes that could not be matched were
Histologic Findings
benign at histologic evaluation. Of the Postcontrast Imaging
172 matching nodes, 116 of the nodes Results Positive Negative All Nodes
that were nonmetastatic at histologic Positive 48 6 54
evaluation showed a signal intensity de- Negative 2 116 118
crease on the ferumoxtran-10 – enhanced Total 50 122 172
MR images (Fig 3). These were true-nega-
Note.—Positive results indicate metastatic nodes, and negative results indicate nonmetastatic
tive nodes on ferumoxtran-10 – enhanced nodes. For postcontrast MR imaging, accuracy was 95%; sensitivity, 96%; specificity, 95%; neg-
MR images. Of the 50 nodes that were ative predictive value, 98%; and positive predictive value, 89%.
metastatic at histologic evaluation, 48
showed no signal intensity decrease after
ferumoxtran-10 administration (true-
positive), and the other two showed a
decrease in signal intensity (false-nega-
tive).
The accuracy, sensitivity, specificity,
negative predictive value, and positive
predictive value of precontrast MR imag-
ing were 92%, 76%, 97%, 91%, and 97%,
respectively (Table 2); corresponding val-
ues for postcontrast MR imaging were
95%, 96%, 95%, 98%, and 89% (Table 3).
The sensitivity and negative predictive
value of the postcotrast imaging were sig-
nificantly better compared with those of
the precontrast imaging (P ⬍ .001 for
sensitivity, P ⬍ .01 for negative predictive
value); however, the change in specificity
Figure 3. T2*-weighted MR images (800/25.4, 30° flip angle, 3.0-mm section thickness) ob-
was not significant (P ⬎ .01).
tained in a plane parallel to psoas muscle. Precontrast image (left) shows normal-sized (7 ⫻ 3 mm)
Of the 50 metastatic lymph nodes, 38 node (ellipse). On postcontrast image (right), this node (circle) shows homogeneous signal
were enlarged (mean size, 15.3 mm; intensity decrease, and another normal-sized (7 ⫻ 4 mm) node with low signal intensity is visible
range, 10 –28 mm) on precontrast MR im- (ellipse). White area is a lymphocele. Histopathologic evaluation confirmed nonmetastatic nodes.
ages and 12 were normal in size (mean,
7.2 mm; range, 6 –9 mm). Ten of these 12
normal-sized nodes showed a lack of
ferumoxtran-10 uptake and signal inten- images (Fig 6). Six (5%) normal-sized these nodes were proved to be metastatic
sity decrease in focal areas on postcon- nodes were false-positive on these post- at histologic evaluation.
trast images, which suggested metastasis contrast MR images. Finally, extracapsular-spread disease was
(Figs 4, 5). Of the 122 benign nodes, one In nine patients, ferumoxtran-10 – en- correctly diagnosed at ferumoxtran-10–
node was false-positive on precontrast hanced MR imaging allowed metastatic enhanced MR imaging in another patient
MR images. This node was enlarged ac- nodes to be found outside of the surgical (Fig 8). Although the node was slightly ir-
cording to size criteria but was correctly field (Fig 7). In two of these nine patients, regular on precontrast images, a clearly reg-
classified as negative on the basis its low the metastatic nodes were normal-sized ular surface was seen at the low signal in-
signal intensity on the postcontrast MR benign nodes on precontrast images; tensity (normal) part and an irregular area

452 䡠 Radiology 䡠 November 2004 Deserno et al

 therapy. However, more extensive nodal
dissection may expose the patient to in-
creased procedure-related complications.
Provision of accurate nodal informa-
tion prior to surgery, by using noninva-
Radiology

sive imaging modalities such as CT, MR

imaging, or PET, may help to reduce
these complications and contribute to a
more complete lymph node dissection.
Nodal assessment with CT or MR imag-
ing relies on insensitive nodal size criteria
and does not have the desired sensitivity
for identification of metastases. The
rather low sensitivity (76%) of nonen-
hanced MR imaging, as found in our
study, concurs with existing results in the
literature (12,13). The reason for this low
Figure 4. T2*-weighted MR images (800/25.4, 30° flip angle, 3.0-mm section thickness) obtained sensitivity is a reliance on insensitive
in a plane parallel to psoas muscle. Precontrast image (left) shows enlarged (17 ⫻ 12 mm) node
nodal size criteria and a lack of differ-
(ellipse). On postcontrast image (right), this same node shows homogeneous signal intensity
decrease. A triangular area with low signal intensity is visible in caudal part of the node, which ences in signal intensity between normal
had high signal intensity on the precontrast image (this is normal nodal tissue). Cranial to this and metastatic lymph nodes (14,15). Al-
area, three more or less round high-signal-intensity metastatic areas (arrows) are visible; they are though promising in the evaluation of
separated by thin strands of low-signal-intensity normal tissue. Results of biopsy in cranial part lung cancer, the role of fluorine 18 flu-
of the node confirmed metastases. orodeoxyglucose PET in the urinary tract
region is limited (4). In a recent study of
PET conducted in 64 patients with blad-
der cancer, Bachor et al (16) found a sen-
sitivity of 67% and a negative predictive
value of 84%.
Published results of previous studies
have shown that ferumoxtran 10 – en-
hanced MR imaging can improve the ac-
curacy for characterizing lymph nodes
(17–22). After intravenous injection, the
iron nanoparticles are transported to nor-
mal-functioning nodal macrophages and
reduce the signal intensity of the nodes
in which they accumulate; this is because
of the T2* and susceptibility effects of the
iron oxide. Normal and benign nodes
thus show negative enhancement (re-
duced signal intensity) after the adminis-
tration of ferumoxtran-10 (Figs 1, 4). In
metastatic nodes, cancer cells replace the
normal-functioning macrophages. Be-
cause the cancer cells lack reticuloendo-
thelial activity and are unable to retain
Figure 5. T2*-weighted MR images (800/25.4, 30° flip angle, 3.0-mm section thickness) in a the iron particles, these nodes retain
plane parallel to psoas muscle. Precontrast image (left) shows normal-sized (12 ⫻ 8 mm) node their high signal intensity after ferumox-
(ellipse). On postcontrast image (right), this same node shows signal intensity decrease with a
tran-10 administration (Figs 4, 5).
small 2-mm area (arrow) of persistant high signal intensity. Histopathologic results confirmed
small 2-mm metastases. Our results confirm the findings of
Bellin et al (17), who determined that
ferumoxtran-10 aids in differentiating
metastatic and benign nodes because of
was seen at the slightly increased signal effect on prognosis in patients with mus- the decreased signal intensity in normal
intensity part on the postcontrast images. cle-invasive bladder cancer (9); they en- lymph nodes. Bellin et al reported a sen-
This increased the suspicion of extracapsu- able identification of those patients who sitivity of 82%, a specificity of 83%, and a
lar spread. require adjuvant chemotherapy. Herr et negative predictive value of 87% for feru-
al (10,11) suggested that a wide extravesi- moxtran-10 – enhanced imaging. Our
DISCUSSION cal soft-tissue margin with bilateral pel- study results showed a higher sensitivity
vic nodal dissection of at least nine nodes (96%), specificity (95%), and negative
Extensive pelvic lymphadenectomy and is required to obtain individual prognos- predictive value (98%); this difference is
pathologic nodal assessment at the time tic information and identify patients probably attributed to our use of high-
of radical cystectomy have a substantial who will benefit from adjuvant chemo- spatial-resolution MR imaging tech-

Volume 233 䡠 Number 2 Preoperative Nodal Staging of Urinary Bladder Cancer 䡠 453
 niques and reduced respiratory and mo-
tion artifacts (13). Also, the higher
specificity in our study can be explained
by our use of a higher dose of contrast
agent (2.6 vs 1.7 mg Fe per kilogram of
Radiology

body weight). In a study in which four

doses were compared (1.1, 1.7, 2.6, and
3.4 mg Fe per kilogram) in 24 healthy
adults, Hudgins et al (23) showed that
the decrease in lymph node signal inten-
sity on ferumoxtran-10 – enhanced MR
images was more pronounced with use of
the dose of 2.6 mg Fe per kilogram than
with the dose of 1.7 mg Fe per kilogram
(as used by Bellin et al).
The high sensitivity and specificity in
our study confirm the results of previous
studies, in patients with other types of Figure 6. T2*-weighted MR images (800/25.4, 30° flip angle, 3.0-mm section thickness) obtained
cancer, in which high-spatial-resolution in a plane parallel to psoas muscle. Precontrast image (left) shows enlarged (14 ⫻ 10 mm) node
(ellipse). On postcontrast image (right), this same node shows homogeneous signal intensity
techniques and a dose of 2.6 mg Fe per
decrease, which indicates absence of malignancy. Histopathologic results confirmed an enlarged
kilogram were also used (7,24 –27). The hyperplastic node.
results are in agreement with the findings
of Jager et al (2), who determined that
malignant nodes in urinary bladder can-
cer are larger, on average, than those
found in prostate cancer. This explains
the higher sensitivity (76%) of nonen-
hanced MR imaging in the urinary blad-
der in our study in comparison with the
results of Harisinghani et al (7) in the
prostate. Nonetheless, in our study, the
sensitivity of ferumoxtran-10 – enhanced
MR imaging was still significantly higher
than that of nonenhanced MR imaging
because of the detection of metastases in
10 small (⬍8 –10-mm) nodes. Sensitivity
of ferumoxtran-10 – enhanced MR imag-
ing in these two cancers, however, is
comparable: 91% in prostate cancer and
96% in bladder cancer.
In our study, one enlarged hyperplastic Figure 7. T2*-weighted MR images (800/25.4, 30° flip angle, 3.0-mm section thickness) in a
node was correctly classified as nonmeta- plane parallel to psoas muscle. Precontrast image (right) shows normal-sized (8 ⫻ 7 mm) node
static on the basis of its low signal inten- (ellipse). On postcontrast image (right), this same node shows homogeneous persistant high
signal intensity. Node is located in the internal iliac region, which is outside the surgical field of
sity on postcontrast MR images. How-
view. On the basis of MR images, lymph node dissection was extended and histopathologic
ever, specificity was decreased due to six results confirmed a normal-sized metastatic node.
heterogeneous nodes with small focal ar-
eas with lack of signal intensity decrease.
In two cases, this decrease could be ex-
plained as being caused by focal fibrosis, “noise” or heterogeneous hilar fat as bly have no negative effect on the out-
as follows: The signal intensity of fibrosis small metastastic foci or reactive hyper- come.
was not different enough from that of plasia was the reason for absence of sig- Our results show a higher sensitivity
normal nodal tissue to be visualized on nal intensity decrease (28). than specificity at ferumoxtran-10 – en-
our T1- and T2*-weighted MR images Earlier studies have addressed the lim- hanced MR imaging. This may not be a
and, therefore, could not be recognized itation of making precise nodal compar- deterrent in a clinical setting; since a ma-
on the precontrast images. Because of the isons between surgery and MR imaging lignant node on ferumoxtran-10 – en-
selective strong decreases in signal inten- (29). To minimize this in our study, we hanced MR images always needs to be
sity of normal nodal tissue on postcon- used a stringent nodal localization proto- histologically confirmed, the only conse-
trast images, fibrosis became visible. In col prior to and during surgery. Nonethe- quence may be the removal of a few ad-
another case that was false-positive at MR less, despite the special care taken in ditional nodes by the urologist. The im-
imaging, histologic results showed focal matching the nodes on MR images with provement in sensitivity from 76% at
lipomatosis. In the remaining cases, its pathologic results, definitely matching precontrast imaging to 96% at postcon-
remains uncertain why the result was pairs were found in only 172 (43%) of trast imaging (P ⬍ .001) implies better
false-positive. One possible explanation 404 nodes. Since all noncomparable visualization of metastatic nodes. There-
may be that the misinterpretation of nodes were benign, this bias will proba- fore, ferumoxtran-10 – enhanced MR im-

454 䡠 Radiology 䡠 November 2004 Deserno et al


Radiology
Figure 8. Transverse T1*-weighted MR images (11.4/4.4, 8° flip angle, 1.4-mm section thick-
ness). Precontrast image (left) shows normal-sized (15 ⫻ 9 mm) node (ellipse). On postcontrast
image (right), this same node shows partial signal intensity decrease (ventral side) and irregular
spread of higher signal intensity tissue in perinodal fat (arrows). Histopathologic results con-
firmed a partially metastatic node with extranodal tumor spread.
aging can better guide the surgeon in dis-
section, which results in the removal of
malignant nodes only. This is clinically
important, as some authors claim that
removing all possible malignant lymph
nodes results in a better prognosis for
local lymph node recurrence a better
probability of patient survival (11,24,30).
Furthermore, it is important to show all
potential locations of malignant nodes
with a higher sensitivity because if a pa-
tient has more than four malignant
lymph nodes or has lymph nodes with
extracapsular tumor growth, cystectomy
is not considered to be the correct option
for improved survival (31). In our study,
information provided by ferumoxtran-
10 – enhanced MR imaging in nine pa-
tients allowed the surgeon to resect
nodes outside the normal surgical dissec-
tion field. In addition, extranodal growth
was visualized in another patient. To
prove that ferumoxtran-10 – enhanced
MR imaging improves the outcome of
lymph node dissection, prospective ran-
domized clinical trials are needed; how-
ever, such studies are not ethically feasi-
ble.
A practical limitation on the use of
ferumoxtran-10 is that it requires a 30-
minute administration period with med-
ical supervision and two separate MR ex-
aminations performed within 24 –36
hours; therefore, use of ferumoxtran-10
is time-consuming and adds expenses.
This problem may be solved in the fu-
ture, since initial reports have shown
that ferumoxtran-10 – enhanced MR im-
aging alone may be sufficient for charac-
terizing the nodes (32). Another limita-
Volume 233 䡠 Number 2
tion of our study was that there was a
single reading, and, therefore, no infor-
mation about interobserver variability
was obtained. However, the images were
read by three independent and experi-
enced readers, which decreases the
chance of interobserver bias. Nonethe-
less, it can be expected that results may
differ when inexperienced readers are
used. Finally, there are probably some
clustering effects on several levels (hospi-
tal, physician, and patient levels). But
without the correction on clustering, the
differences between pre- and postcon-
trast MR imaging were so great at logistic
regression analysis that we did not think
it was necessary to correct for clustering
effects. We did not distinguish between
different hospitals because the nodal
numbers did not allow this.
At ferumoxtran-10 – enhanced MR im-
aging, normal nodal tissue and metasta-
sis have different signal intensities; this
difference allows detection of metastases
even in normal-sized nodes, which re-
sults in increased sensitivity, from 76% to
96%, and negative predictive value, from
91% to 98%, in comparison with the sen-
sitivity of nonenhanced MR imaging.
The use of this technique thus provides a
better detection of metastatic nodes,
which can improve the accuracy of sur-
gical nodal resection.
References
1. Barentsz JO, Jager GJ, Witjes JA. MR im-
aging of the urinary bladder. Magn Reson
Imaging Clin N Am 2000; 8:853– 867.
2. Jager GJ, Barentsz JO, Oosterhof G, et al.
Pelvic adenopathy in prostatic and uri-
nary bladder carcinoma: MR imaging
Preoperative Nodal Staging of Urinary Bladder Cancer
with a three-dimensional T1-weighted
magnetization-prepared rapid gradient-
echo sequence. AJR Am J Roentgenol
1996; 167:1503–1507.
3. Barentsz JO, Jager GJ, van Vierzen PB, et
al. Staging urinary bladder cancer after
transurethral biopsy: value of fast dy-
namic contrast-enhanced MR imaging.
Radiology 1996; 201:185–193.
4. Pieterman RM, van Putten JW, Meuzelaar
JJ, et al. Preoperative staging of non-
small-cell lung cancer with positron-
emission tomography. N Engl J Med
2000; 343:254 –261.
5. Bellin MF. Iron oxide-enhanced MR lym-
phography: initial experience. Eur J Ra-
diol 2000; 34:257–264.
6. Harisinghani MG, Saini S, Weissleder R, et
al. MR lymphangiography using USPIO in
patients with primary abdominal and pel-
vic malignancies: radiographic-patho-
logic correlation. AJR Am J Roentgenol
1999; 172:1347–1351.
7. Harisinghani M, Barentsz J, Hahn P, et al.
Non-invasive detection of clinically oc-
cult lymph-node metastasis in prostate
cancer. N Engl J Med 2003; 348:2491–
2499.
8. Sobin LH, Wittekind CH, eds. Interna-
tional Union Against Cancer (UICC):
TNM classification of malignant tumors.
New York, NY: Wiley-Liss, 1997; 170 –
173.
9. Stein JP, Lieskovsky G, Cote R, et al. Rad-
ical cystectomy in the treatment of inva-
sive bladder cancer: long-term results in
1,054 patients. J Clin Oncol 2001;
19:666 – 675.
10. Herr HW, Bochner BH, Dalbagni G, et al.
Impact of the number of lymph nodes
retrieved on outcome in patients with
muscle invasive bladder cancer. J Urol
2002; 167:1295–1298.
11. Herr HW. Extent of surgery and pathol-
ogy evaluation has an impact on bladder
cancer outcomes after radical cystectomy.
Urology 2003; 61:105–108.
12. Wolf JS, Cher M, Dall’era M, et al. The use
and accuracy of cross-sectional imaging
and fine needle aspiration cytology for
detection of pelvic lymph node metasta-
ses before radical prostatectomy. J Urol
1995; 153:993–999.
13. Scheidler J, Hricak H, Yu KK, et al. Radio-
logical evaluation of lymph node metas-
tases in patients with cervical cancer.
JAMA 1997; 278:1096 –1101.
14. Dooms GC, Hricak H, Crooks LE, et al.
Magnetic resonance imaging of the
lymph nodes: comparison with CT. Radi-
ology 1984; 153:719 –728.
15. Dooms GC, Hricak H, Mosely MR, et al.
Characterization of lymphadenopathy by
magnetic relaxation times: preliminary
results. Radiology 1985; 155:691– 697.
16. Bachor R, Kotzerke J, Reske SN, et al.
Lymph node staging of bladder neck car-
cinoma with positron emission tomogra-
phy. Urologe 1999; 38:46 –50.
17. Bellin MF, Roy C, Kinkel K, et al. Lymph
node metastases: safety and effectiveness
of MR imaging with ultrasmall superpara-
manetic iron oxide particles—initial clin-
ical experience. Radiology 1998; 207:
799 – 808.
18. Weissleder R, Elizondo G, Wittenberg J, et
al. Ultrasmall paramagnetic iron oxide:
an intravenous contrast agent for assess-
䡠 455
 ing lymph nodes with MR imaging. Radi-
ology 1990; 175:494 – 498.
19. Vassallo P, Matei C, Heston WDW, et al.
AMI-227-enhanced MR lymphography:
usefulness for differentiating reactive
from tumor bearing lymph nodes. Radi-
ology 1994; 193:501–506.
Radiology
20. Gerlowski LE, Jain RK. Microvascular per-
meability of normal and neoplastic tis-
sues. Microvasc Res 1986; 31:288 –305.
21. Guimaraes R, Clement O, Bittoun J, et al.
MR lymphography with super paramag-
netic nanoparticles in rats: pathologic ba-
sis for contrast enhancement. AJR Am J
Roentgenol 1994; 162:201–207.
22. Chambon C, Clement O, Le Blanche A, et
al. Superparamagnetic iron oxides as pos-
itive MR contrast agents: in vitro and in
vivo evidence. Magn Reson Imaging
1993; 11:509 –519.
23. Hudgins PA, Anzai Y, Christino A, et al.
Dextran-coated superparamagnetic iron
oxide MR contrast agent (Combidex) for
imaging cervical lymph nodes: optimal
dose, time of imaging, and pulse se-
456 䡠 Radiology 䡠 November 2004
quence (abstr). Proceedings of the 34th
Annual Meeting of the American Society
of Neuroradiology. Oak Brook, Ill: Amer-
ican Society of Neuroradiology, 1996; 24.
24. Anzai Y, Blackwell K, Hirschowitz S, et al.
Initial clinical experience with dextran-
coated superparamagnetic iron oxide for
detection of lymph node metastases in
patients with head and neck cancer. Ra-
diology 1994; 192:709 –715.
25. Stets C, Brandt S, Wallis F, et al. Axillary
lymph node metastases: a statistical anal-
ysis of various parameters in MRI with
USPIO. J Magn Reson Imaging 2002; 16:
60 – 68.
26. Pannu HK, Wang KP, Borman TL, et al.
MR imaging of mediastinal lymph nodes:
evaluation using a superparamagnetic
contrast agent. J Magn Reson Imaging
2000; 12:899 –904.
27. Michel SC, Keller TM, Frohlich PH, et al.
Preoperative breast cancer staging: MR
imaging of the axilla with ultrasmall su-
perparamagnetic iron oxide enhance-
ment. Radiology 2002; 225:527–536.
28. Bellin MF, Leblue L, Meric JB. Evaluation
of retroperitoneal and pelvic lymph node
metastases with MRI and MR lym-
phangiography. Abdom Imaging 2003;
28:155–163.
29. Reinhardt MJ, Ehritt-Braun C, Vogelsnag
D, et al. Metastatic lymph nodes in pa-
tients with cervical cancer: detection
with MR Imaging and FDG PET. Radiol-
ogy 2001; 218:776 –778.
30. Leissner J, Allhof EP, Hohenfeller R, et al.
Significance of pelvic lymphadenectomy
for the prognosis after radical cystec-
tomy. Zentralbl Chir 2002; 127:315–332.
31. Mills RD. Pelvic lymph node metastases
from bladder cancer outcome in 83 pa-
tients after radical cystectomy and pelvic
lymphadenectomy. J Urol 2001; 166:19 –
23.
32. Harisinghani MG, Barentsz JO, Hahn PF,
et al. MR lymphangiography for detec-
tion of minimal nodal disease in patients
with prostate cancer (abstr). Radiology
2001; 221(P):387.
Deserno et al