AJR:175, September 2000

759

Comparison of Dynamic Helical CT
and Dynamic MR Imaging in the
Evaluation of Pelvic Lymph Nodes
in Cervical Carcinoma

OBJECTIVE

.

This study compares dynamic helical CT with dynamic MR imaging in the
evaluation of pelvic lymph nodes in cervical carcinoma.

SUBJECTS AND METHODS

.

Women with biopsy-proven cervical carcinoma pro-
spectively underwent dynamic helical CT and MR imaging before surgery. A metastatic node
on CT and MR imaging was dened as a rounded soft-tissue structure greater than 10 mm in
maximal axial diameter or a node with central necrosis. Imaging results were compared with
pathology, and receiver operating characteristic curves for size and shape were plotted on a
hemipelvis basis.



Nodal density and signal intensity on CT and MR images, respectively,
were reviewed for differences between benign and malignant disease.

RESULTS

.

A total of 949 lymph nodes were found at pathology in 76 hemipelves in 43
women, of which 69 lymph nodes (7%) in 17 hemipelves (22%) were metastatic. Sensitivity,
specicity, positive and negative predictive values, and accuracy of helical CT and MR imag-
ing in the diagnosis of lymph node metastasis on a hemipelvis basis was 64.7%, 96.6%,
84.6%, 90.5%, and 89.5% and 70.6%, 89.8%, 66.7%, 91.4%, and 85.5%, respectively. Re-
ceiver operating characteristic curves for helical CT and MR imaging gave cutoff values of 9
and 12 mm in maximal axial diameter, respectively, in the prediction of metastasis. Central
necrosis had a positive predictive value of 100% in the diagnosis of metastasis. Signal inten-
sity on MR imaging and densityenhancement pattern on CT in patients with metastatic
nodes did not differ from those in patients with negative nodes.

CONCLUSION

.

Helical CT and MR imaging show similar accuracy in the evaluation of
pelvic lymph nodes in patients with cervical carcinoma. Central necrosis is useful in the diag-
nosis of metastasis in pelvic lymph nodes in cervical cancer.
elvic and paraaortic nodal status
is of utmost importance in the
prognosis of and therapy plan-
ning for patients with cervical carcinoma;
this makes preoperative assessment of nodal
involvement essential.



Various accuracy rates
have been reported for CT and MR imaging
in the detection of pelvic nodal metastases
from uterine carcinoma and other pelvic ma-
lignancies [17]. The validity of criteria
other than nodal size including shape, pres-
ence of central necrosis, inherent tissue con-
trast, and degree of contrast enhancement in
pelvic nodes has not been thoroughly ex-
plored. Several studies have evaluated lymph
nodes using thin-section helical CT in the
neck, thorax, and abdomen [811]. To the
best of our knowledge, no studies on the use
of dynamic helical CT or dynamic MR imag-
ing in the evaluation of pelvic lymph nodes
in cervical carcinoma have been published.
The aim of this study is to determine the ac-
curacy of dynamic enhanced helical CT and
MR imaging in the detection of pelvic nodal
metastases when compared with pathology.

Subjects and Methods

Patient Population

Women with biopsy-proven cervical cancer
who were scheduled for surgery were prospec-
tively recruited for the study to undergo dynamic
helical CT and dynamic MR imaging within 2
weeks before surgery. This study received local in-
stitutional ethics approval, and informed consent
was obtained from all patients.

Helical CT Protocol

Helical CT examinations were performed on a
HiSpeed Advantage scanner (General Electric
Medical Systems, Milwaukee, WI) with the fol-

Wei Tse Yang

1

Wynnie Wai Man Lam

1

Mei Yung Yu

2

Tak Hong Cheung

3

Constantine Metreweli

1

Received November 8, 1999; accepted after revision
February 8, 2000.

1

Department of Diagnostic Radiology and Organ Imaging,
Chinese University of Hong Kong, Prince of Wales Hospital,
Shatin, New Territories, Hong Kong. Address
correspondence to W. T. Yang.

2

Department of Anatomical and Cellular Pathology,
Chinese University of Hong Kong, Prince of Wales Hospital,
Shatin, New Territories, Hong Kong.

3

Department of Obstetrics and Gynecology, Chinese
University of Hong Kong, Prince of Wales Hospital, Shatin,
New Territories, Hong Kong.

AJR

2000;175:759766
0361803X/00/1753759
American Roentgen Ray Society

P
D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
j
r
o
n
l
i
n
e
.
o
r
g

b
y

1
1
4
.
4
.
2
3
.
0

o
n

0
9
/
0
8
/
1
4

f
r
o
m

I
P

a
d
d
r
e
s
s

1
1
4
.
4
.
2
3
.
0
.

C
o
p
y
r
i
g
h
t

A
R
R
S
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
;

a
l
l

r
i
g
h
t
s

r
e
s
e
r
v
e
d


760

AJR:175, September 2000

Yang et al.

lowing protocol. An unenhanced scan was ob-
tained cephalad from the symphysis pubis to the
iliac crest at 7-mm slice thickness, 1.5:1 pitch, and
7-mm reconstruction interval. A bolus of 120 mL
of IV contrast material (iopromide 240 [Ultravist
240]; Schering, Berlin, Germany) was adminis-
tered using an automatic injector: the rst 100 mL
were delivered at a rate of 2.5 mL/sec and the re-
maining 20 mL at 0.5 mL/sec. Data acquisition
was divided into two parts: at 40 sec after the start
of contrast material injection, offering a good arte-
rial phase, and at 180 sec after injection, when
veins are optimally enhanced. Scanning was per-
formed during inspiration using the following pa-
rameters: eld of view, 320340 mm; 120 kV;
250300 mA; matrix, 512

512

; and a soft-tissue
algorithm reconstruction.

MR Imaging Protocol

MR examinations were performed on a 1.5-T
scanner (Gyroscan ACS-NT; Philips Medical Sys-
tems, Best, The Netherlands) using a body coil. All
patients underwent an axial T1-weighted spin-echo
sequence (TR/TE, 550/15; eld of view, 350450
mm; slice thickness, 5 mm with 0.5-mm interslice
gap),



an



axial turbo spin-echo T2-weighted short-tau
inversion recovery sequence (1800/100; inversion
time, 180 msec; eld of view, 350450 mm; slice
thickness, 6 mm with 1.2-mm interslice gap), and a
sagittal turbo spin-echo T2-weighted sequence (2500/
150; eld of view, 300450 mm; slice thickness, 4
mm with 0.4-mm interslice gap). Turbo spin-echo dy-
namic axial scans (500/12 and 4-mm thickness with
no interslice gap) were obtained immediately after
hand injection of 0.1 mmol/kg of gadolinium (Mag-
nevist; Schering) at a rate of 2 mL/sec. A total of six
B A
Fig. 1.31-year-old woman with right hypogastric lymph node metastasis found at pathology.
A, Unenhanced axial CT scan of pelvis shows round mildly hypodense nodule (arrow) in right hypogastric region.
B, Arterial phase axial CT scan obtained at same level as A shows heterogeneous peripheral enhancement of node (solid arrow) and external iliac artery (open arrow)
opacication.
C, Venous phase CT scan obtained at same level as B shows peripheral rim enhancement of node (long black arrow). Prominence of intranodal necrosis exceeds that seen
in A and B. Note iliac vein (short black arrow) enhancement and ureteric (curved white arrow) opacication.
D, Axial short-tau inversion-recovery fat-suppressed T2-weighted MR image obtained at same level as C shows right hypogastric node (solid arrow) with hyperintense
area (open arrow) that was hypointense on T1-weighted image (not shown), representing intranodal necrosis.
D C
D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
j
r
o
n
l
i
n
e
.
o
r
g

b
y

1
1
4
.
4
.
2
3
.
0

o
n

0
9
/
0
8
/
1
4

f
r
o
m

I
P

a
d
d
r
e
s
s

1
1
4
.
4
.
2
3
.
0
.

C
o
p
y
r
i
g
h
t

A
R
R
S
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
;

a
l
l

r
i
g
h
t
s

r
e
s
e
r
v
e
d


Dynamic Imaging of Cervical Carcinoma

AJR:175, September 2000

761

dynamic scans were obtained with an acquisition time
of 3 min 45 sec.

Interpretation of Helical CT and MR Images

Both helical CT and MR images were interpreted
by two radiologists by consensus and without
knowledge of the ndings of the other technique or
of nal pathologic diagnosis. For the interpretation
of helical CT and MR images, identication of
lymph nodes was conducted on the CT and MR con-
soles, with unenhanced arterial and venous phase
slices on helical CT arranged consecutively for the
same level. All helical CT and MR images were re-
corded on hard copy (3M Medical Imaging Sys-
tems, St. Paul, MN) laser images for reevaluation
and cross-reference as necessary. The presence of all
pelvic nodes along the iliac (common, internal, and
external) chains was noted regardless of size. Then,
the following criteria for the diagnosis of metastasis
in pelvic nodes were used: maximal axial diameter
of more than 10 mm and the presence of central ne-
crosis regardless of nodal size. On both unenhanced
and enhanced CT, central necrosis was dened as a
central density of less than 20 H (Figs. 1A1C); on
MR imaging, as an intranodal isointense (to water)
area on both T1- and T2-weighted images (Fig. 1D)
that showed no enhancement after contrast material
administration (Fig. 2). The minimal axial diameter
of each lymph node identied on helical CT and MR
imaging was also documented, and the maximal ax-
ial diameterminimal axial diameter ratio was calcu-
lated. In addition, receiver operating characteristic
(ROC) curves on size and shape (maximal axial di-
ameterminimal axial diameter ratio) were plotted
to verify the accuracy of published cutoff values.

Dynamic Imaging Evaluation

The inherent nodal tissue contrast as determined
by attenuation value and signal intensity on helical
CT and MR imaging, respectively, was assessed on
unenhanced images. Enhancement pattern of the
largest node per anatomic (common, internal, and
external iliac) chain was assessed after contrast-en-
hanced dynamic CT and MR imaging. The attenua-
tion value of each node was compared on the
unenhanced, arterial phase, and venous phase helical
CT scans. For each node visualized on MR imaging,
the signal intensity before and after the administra-
tion of contrast material was noted. Of a total of six
scans, the dynamic scan in which the greatest rate of
enhancement occurred was also noted. In nodes with
an area of low density on CT or water isointensity on
MR imaging, respectively, measurements were
made from the noncystic portion of the node.

Pathology

All patients subsequently underwent pelvic
lymph node dissection with anatomic labeling into
common, internal, and external iliac groups with
evaluation by a single pathologist. The total number
of lymph nodes, number of metastatic nodes, and
maximum size of the largest metastatic lymph node
were documented. The imaging ndings were com-
pared with histology, and the accuracy of each tech-
nique in each hemipelvis was determined.

Statistics

For the purpose of calculations of accuracy and
ROC analysis of nodal size and shape, ndings
were considered true-positive on CT or MR imag-
ing if nodes in a hemipelvis met the size criterion
or showed the presence of central necrosis (or
both) and if at least one positive node was found at
dissection anywhere in that hemipelvis. Findings
were considered false-positive on CT or MR imag-
ing if nodes in a hemipelvis met the size criterion
or showed the presence of central necrosis (or
both) and no metastatic node was found at dissec-
tion anywhere in that hemipelvis. Findings were
considered false-negative on CT or MR imaging if
nodes in a hemipelvis did not meet the size crite-
rion or did not show the presence of central necro-
sis (or both) and if a metastatic node was found at
dissection anywhere in that hemipelvis. Findings
were considered true-negative on CT or MR imag-
ing if nodes in a hemipelvis did not meet the size
criterion or did not show the presence of central
necrosis (or both) and if no metastatic node was
found at dissection anywhere in that hemipelvis.

Results

Pathology

A total of 43 women (age range, 2179
years; mean age, 46 years) who underwent sur-
gery with pelvic lymph node dissection formed
the study cohort. According to the criteria of the
International Federation of Gynecology and
B A
Fig. 2.39-year-old woman with left external iliac nodal metastasis at pathology.
A, Unenhanced axial T1-weighted MR image shows ovoid hypointense nodule (arrow) in left external iliac region.
B, Dynamic contrast-enhanced MR image obtained at same level as A shows heterogeneous peripheral enhancement (arrows) with nonenhancing central necrotic region
(asterisk).
D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
j
r
o
n
l
i
n
e
.
o
r
g

b
y

1
1
4
.
4
.
2
3
.
0

o
n

0
9
/
0
8
/
1
4

f
r
o
m

I
P

a
d
d
r
e
s
s

1
1
4
.
4
.
2
3
.
0
.

C
o
p
y
r
i
g
h
t

A
R
R
S
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
;

a
l
l

r
i
g
h
t
s

r
e
s
e
r
v
e
d


762

AJR:175, September 2000

Yang et al.

Obstetrics (FIGO) [12], two women had clini-
cal stage 1a disease, 29 had stage 1b, 10 had
stage 2a, and two had stage 2b. There were 29
squamous cell carcinomas, 12 adenocarcino-
mas, and one clear cell and one mixed type car-
cinoma (adenocarcinoma and small cell
carcinoma) each. Pelvic lymphadenectomy on
76 hemipelves yielded a total of 949 lymph
nodes (range, 141 nodes per patient; mean, 22
nodes per patient). Total pelvic lymphadenec-
tomy was performed in 33 patients, and unilat-
eral pelvic lymphadenectomy in 10 women
who had frozen-section conrmation of me-
tastasis in clinically suspicious pelvic lymph
nodes at surgery. All patients with metastatic
nodes required further treatment with radiation
therapy at our institution. As a result, surgeons
do not perform radical surgery (radical hyster-
ectomy and total pelvic lymphadenectomy) in
the presence of nodal metastasis in an effort to
decrease morbidity from combined radical sur-
gery and postoperative radiation therapy. Me-
tastasis was conrmed pathologically in 17
hemipelves (22%). Sixty-nine lymph nodes
(7%) in 15 women (range, 111 nodes per pa-
tient; mean, 4.6 nodes per patient) were meta-
static at pathology.

Accuracy Using Standard Published Criteria for Size

Helical CT revealed a total of 108 lymph
nodes, of which 26 (24%) were diagnosed as
metastatic. MR imaging revealed a total of
380 nodes, of which 40 (11%)



were diag-
nosed as metastatic. Therefore, 76% (82/
108) of the nodes identied on CT had a
maximal axial diameter of less than 10 mm,
whereas 90% (340/380) of the nodes seen on
MR imaging had a maximal axial diameter
of less than 10 mm. The results of helical CT
diagnosis of metastatic lymph nodes on a
hemipelvis basis were as follows: sensitivity,
specicity, positive and negative predictive
values, and accuracy were 64.7%, 96.6%,
84.6%, 90.5%, and 89.5% compared with
70.6%, 89.8%, 66.7%, 91.4%, and 85.5%,
respectively, for MR imaging (Table 1).
There were two false-positive and six false-
negative ndings each using helical CT, and
six false-positive and ve false-negative nd-
ings using MR imaging (Tables 2 and 3).

ROC Verication

Size

.The ROC curves for size on helical
CT and MR imaging are shown in Figures 3
and 4. Optimal maximal axial diameter and
minimal axial diameter for helical CT was 9
and 8 mm, respectively, with accuracy pro-
les (sensitivity, specicity, positive and neg-
ative predictive values, and accuracy) of
70.6%, 85.7%, 66.7%, 87.8%, and 81.4% and
64.7%, 81.0%, 57.9%, 85.0%, and 76.3%, re-
spectively (Table 4). Optimal maximal axial
diameter and minimal axial diameter for MR
imaging was 12 and 8 mm, respectively, with
accuracy proles (sensitivity, specicity, pos-
itive and negative predictive values, and accu-
racy) of 82.4%, 79.3%, 53.9%, 93.9%, and
80.0% and 76.5%, 81.0%, 54.2%, 92.2%, and
80.0%, respectively (Table 5).

Shape

.The optimal maximal axial di-
ameterminimal axial diameter ratio using
both helical CT and MR imaging was 1.3
and gave sensitivity, specicity, positive and
negative predictive values, and accuracy of
41.2%, 85.7%, 53.9%, 78.3%, and 72.9%
and 47.1%, 86.2%, 50.0%, 84.8%, and
77.3%, respectively (Table 6).

Central Necrosis

Central necrosis was seen on CT in seven
(27%)



of 26 and on MR imaging in seven
(17.5%) of 40 abnormal lymph nodes, all of
which were pathologically metastatic. These
nodes were seen in a total of nine hemipelves,
all of which had metastasis at pathology.
TABLE 1
Comparison of Pathology
with CT and MR Imaging in
the Diagnosis of Metastatic
Pelvic Lymph Nodes by
Hemipelvis
Imaging
Findings
Pathology
Total
Metastasis Benign
CT
Metastasis 11 2 13
Benign 6 57 63
Total 17 59 76
MR imaging
Metastasis 12 6 18
Benign 5 53 58
Total 17 59 76
a
Three hemipelves with false-negative findings on CT had nodes that measured more than 1.0 cm.
b
Three hemipelves with false-negative findings on CT had metastases in the contralateral hemipelvis, and two false-negative hemipelves on MR imaging had metastases in the contralateral
hemipelvis.
c
Two hemipelves had false-negative findings on both CT and MR imaging. Maximum number of false-negative nodes per hemipelvis was 11 using helical CT and nine using MR imaging.
d
Two hemipelves with false-negative findings on MR imaging had nodes that measured more than 1.0 cm.
TABLE 2 False-Negative Findings on CT and MR Imaging of Lymph Nodes in Hemipelves
Imaging Technique Case No.
Imaging Pathology
Total No. of Nodes
Seen
Size of Largest Node
(cm)
No. of Metastatic Nodes/Total No. of Nodes
Size of Largest
Metastatic Node (cm)
Per Hemipelvis Per Patient
CT
a
1 1 0.8 7/12 7/23
b
1.8
2 1 0.8 2/10 3/22 1.0
3 2 0.8 11/21 11/33 1.2
4 1 0.7 1/13 1/13 1.5
5
c
2 0.6 2/7 2/20
b
0.9
6
c
1 0.7 1/10 1/25
b
0.8
MR imaging
d
1 7 0.9 9/23 10/26 1.5
2 4 0.7 1/10 3/20 1.5
3 2 0.8 2/10 3/20 1.0
4
c
7 0.5 2/7 2/20
b
0.9
5
c
5 0.7 1/10 1/25
b
0.8
D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
j
r
o
n
l
i
n
e
.
o
r
g

b
y

1
1
4
.
4
.
2
3
.
0

o
n

0
9
/
0
8
/
1
4

f
r
o
m

I
P

a
d
d
r
e
s
s

1
1
4
.
4
.
2
3
.
0
.

C
o
p
y
r
i
g
h
t

A
R
R
S
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
;

a
l
l

r
i
g
h
t
s

r
e
s
e
r
v
e
d


Dynamic Imaging of Cervical Carcinoma

AJR:175, September 2000

763

Therefore, the presence of central necrosis
had a positive predictive value of 100% in the
diagnosis of metastasis. Pathologic conrma-
tion of necrosis in metastatic nodes from the
anatomic chain corresponding to imaging
was available in ve nodes. The mean size
(maximal axial diameter) of nodes with cen-
tral necrosis on CT was 2.3 cm, with six of
seven nodes measuring more than 2 cm. The
mean size (maximal axial diameter) of nodes
with central necrosis on MR imaging was 1.9
cm with four of seven nodes measuring more
than 2 cm. Five nodes with necrosis seen on
CT or MR imaging (or both) had conrma-
tion of nodal necrosis in a metastatic node
from the same anatomic chain at pathology.
Two nodes diagnosed as necrotic on CT
(maximal axial diameter of 2.5 and 2.1 cm)
were enlarged but not necrotic on MR imag-
ing. Similarly, two necrotic nodes that were
seen on MR imaging (maximal axial diame-
ter of 2 and 1.5 cm) were enlarged on CT but
did not appear necrotic.

Signal Intensity

The scatterplot of signal intensity mea-
surements obtained from 146 lymph nodes is
shown in Figure 5. The mean unenhanced
signal intensity of nodes from patients with
metastasis at pathology (704 224) was sig-
nicantly different compared with that of
nodes from patients with benign pathology
(833 252) (

p

= 0.01). However, the en-
hanced signal intensity of nodes from the
metastatic (1101 260) and benign (1143
278) groups of patients showed no signicant
difference (

p

= 0.4). Likewise, the difference
in unenhanced and enhanced intensity for
benign and malignant groups was not signi-
cant (

p

= 0.09).
Analysis of the ranking of peak rate or
highest gradient of enhancement out of six
dynamic MR images showed no signicant
difference between benign and malignant
groups (Wilcoxons rank sum test,

p

= 0.5),
with the mean ranking for both groups ob-
tained on the third dynamic scan (Fig. 6).
Note.All nodes showed sinus histiocytosis at pathology. One hemipelvis with a false-positive finding on CT had metastasis
in the contralateral hemipelvis, and two hemipelves with false-positive findings on MR imaging had metastasis in the contralat-
eral hemipelvis at pathology.
a
One hemipelvis had false-positive findings on both CT and MR imaging.
b
The discrepancy in nodal size (case 1) on CT and pathology is likely related to measurement of two conglomerate nodes on
CT. MR imaging showed two enlarged nodes adjacent to one another in the same case, whereas three enlarged nodes were
seen at pathology in the same nodal chain.
TABLE 3
False-Positive Findings on CT and MR Imaging of Lymph Nodes Per
Hemipelvis
Imaging Technique Case No.
No. of False-Positive
Nodes Per Hemipelvis
Size (cm) of Largest
False-Positive Node
at Imaging
Node at Pathology
CT 1
a
1 2.5
b
1.2
2 1 1.2 1.1
MR imaging 1
a
1 1.5 1.2
2 1 1.8 1.2
3 1 1.8 1.3
4 2 1.5 1.1
5 1 2.5 3.8
6 2 1.5 1.4
Fig. 3.Receiving operating characteristic curves of maximal axial diameter
obtained with CT (dotted line) and with MR imaging (solid line) show area un-
der curve to be 85.7% and 83.5%, respectively.
Fig. 4.Receiving operating characteristic curves of minimal axial diameter obtained
with CT (dotted line) and with MR imaging (solid line) show area under curve to be 81.7%
and 86.8%, respectively.
D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
j
r
o
n
l
i
n
e
.
o
r
g

b
y

1
1
4
.
4
.
2
3
.
0

o
n

0
9
/
0
8
/
1
4

f
r
o
m

I
P

a
d
d
r
e
s
s

1
1
4
.
4
.
2
3
.
0
.

C
o
p
y
r
i
g
h
t

A
R
R
S
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
;

a
l
l

r
i
g
h
t
s

r
e
s
e
r
v
e
d


764

AJR:175, September 2000

Yang et al.

Dynamic Helical CT

All nodes measured in this study showed en-
hancement after administration of contrast ma-
terial. The mean nodal attenuation value



from
the benign group was 37.2 H (95% condence
interval [CI], 28.046.4) on unenhanced im-
ages, increasing to 59.9 H (95% CI, 43.476.4)
on the arterial phase images, and to 65.6 H
(95% CI, 45.985.3) on the venous phase im-
ages. For the metastatic group, the mean nodal
attenuation value



was 35.5 H (95% CI, 16.4
54.6) on unenhanced images, increasing to 61.5
H (95% CI, 23.099.9) on the arterial phase
images, and to 64.5 H (95% CI, 34.194.9) on
the venous phase images. No difference in at-
tenuation value was noted between nodes from
patients with benign and malignant disease on
all three (unenhanced, arterial, and venous) sets
of dynamic images (

p

> 0.05).

Discussion

Although not included in the FIGO clini-
cal staging, pelvic lymph node status is one
of the most important prognostic factors in
cervical carcinoma. The techniques for de-
tection of pelvic lymph node metastasis in-
clude lymphography, CT, and MR imaging
[18, 13, 14]. These methods each have limi-
tations, the most important being the dif-
culty in differentiating metastatic from
nonmetastatic reactive nodes of similar size.
With the development of helical CT and the
availability of dynamic scanning with both
CT and MR imaging, we were interested in
knowing whether the addition of dynamic
scans would enhance the diagnostic capabil-
ity of helical CT or MR imaging.

Number of Nodes Identied

MR imaging identied a signicantly
greater overall number of lymph nodes (

n

=
380) compared with helical CT (

n

= 108). Most
MR imagingdetected nodes (90%) had a max-
imal axial diameter of 1 cm or less, compared
with 76% of nodes with a maximal axial diam-
eter of 1 cm or less when using CT. A possible
reason for this discrepancy is likely related to
the slice thickness used with CT and MR imag-
ing. Axial MR images used a slice thickness of
4 mm compared with an effective slice thick-
ness of 10.5 mm on helical CT (7-mm slice
thickness and 1.5:1 pitch). The higher resolu-
tionrelated to lower slice thickness on MR
imagingwould enable greater visualization
of smaller nodes. The scanning parameters for
the helical CT protocol were considered opti-
mal considering the scanning time, tube cool-
ing, and radiation dose for dynamic arterial and
venous phase scanning. These parameters will
likely improve with newer generation scanners,
which will have a much smaller effective colli-
mation (almost 5 mm) and will not be limited
by concerns for tube cooling and scanning
TABLE 4
Comparison of Pathology
with Maximal and Minimal
Axial Diameters on CT by
Hemipelvis
Axial Diameter
of Lymph Node
Pathology
Total
Metastasis Benign
Maximal
9 mm 12 6 18
<9 mm 5 36 41
Total 17 42 59
Minimal
8 mm 11 8 19
<8 mm 6 34 40
Total 17 42 59
TABLE 5
Comparison of Pathology
with Maximal and Minimal
Axial Diameters on MR
Imaging by Hemipelvis
Axial Diameter
of Lymph Node
Pathology
Total
Metastasis Benign
Maximal
12 mm 14 12 26
<12 mm 3 46 49
Total 17 58 75
Minimal
8 mm 13 11 24
<8 mm 4 47 51
Total 17 58 75
TABLE 6
Comparison of Pathology
with MaximalMinimal Axial
Diameter Ratio Using CT
and MR Imaging by
Hemipelvis
MaximalMinimal
Axial Diameter Ratio
Pathology
Total
Metastasis Benign
CT
1.3 7 6 13
<1.3 10 36 46
Total 17 42 59
MR imaging
1.3 8 8 16
<1.3 9 50 59
Total 17 58 75
Fig. 5.Scatterplot of unenhanced
signal intensity of pelvic nodes mea-
sured in patients with benign and in
those with malignant pelvic lymph
nodes at histology. Although mean
value of unenhanced signal intensity
for malignant nodes (704 224) is
lower than that for benign nodes
(833 252), signicant overlap exists.
D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
j
r
o
n
l
i
n
e
.
o
r
g

b
y

1
1
4
.
4
.
2
3
.
0

o
n

0
9
/
0
8
/
1
4

f
r
o
m

I
P

a
d
d
r
e
s
s

1
1
4
.
4
.
2
3
.
0
.

C
o
p
y
r
i
g
h
t

A
R
R
S
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
;

a
l
l

r
i
g
h
t
s

r
e
s
e
r
v
e
d


Dynamic Imaging of Cervical Carcinoma

AJR:175, September 2000

765

time. Further reasons that may explain the dis-
crepancy in visibility of the number of lymph
nodes include the ability of MR imaging to dif-
ferentiate blood vessels from lymph nodes and
the ability of multiplanar MR sequences to of-
fer three-dimensional evaluation.

Size and Shape

The only generally accepted CT and MR
imaging criterion in the diagnosis of meta-
static pelvic lymphadenopathy is size [15
18]. Diameter limits ranging from 6 to 15
mm have been used, with 10 mm being the
most commonly used criterion for the upper
limit of a normal lymph node [1518]. A
minimal axial diameter of more than 11.5
cm has been found to be the most valid crite-
rion for detection of metastatic cervical
lymph nodes [19]. The highest reported ac-
curacy rate of 93%, to date, for MR diagno-
sis of metastatic pelvic nodes in cervical
cancer was achieved using a minimal axial
diameter of more than 10 mm [1]. In this
study, both maximal axial diameter and min-
imal axial diameter were equal in diagnostic
accuracy, with optimal values of 9 and 8 mm,
respectively, using helical CT and 12 and 8
mm, respectively, using MR imaging.
It is accepted that neither MR imaging
nor helical CT can depict small metastatic
deposits in normal-sized nodes [18]. Addi-
tional criteria besides diameter include
asymmetry and a rounded or spheric shape.
Knowledge of the conguration of lymph
nodes would be useful in the evaluation of
relatively large (>10 mm) lymph nodes. The
maximal axial diameterminimal axial di-
ameter ratio using both CT and MR imaging
showed poor sensitivity in this study. This
result may be explained by the fact that axial
CT and MR images are not likely to show
the maximum cross section of lymph nodes
compared with other imaging techniques
such as sonography.

Central Necrosis

Central necrosis was seen in 27% and
17% of all abnormal nodes in helical CT and
MR imaging, respectively, and had a positive
predictive value of 100% for metastasis in
this study. Most of the necrotic nodes had a
maximal axial diameter of greater than 2 cm,
although necrotic nodes were generally
smaller on MR imaging (mean maximal ax-
ial diameter, 1.9 cm) than on CT (mean max-
imal axial diameter, 2.3 cm). Two necrotic
nodes that were seen on CT did not appear
necrotic on MR imaging. It is possible that
the presence of proteinaceous uid within
these nodes resulted in a signal intensity that
was less isointense to water. Also, two ne-
crotic nodes that were seen on MR imaging
did not appear necrotic on CT. These areas of
necrosis may have been less evident on CT
because a greater slice thickness was used.
To the best of our knowledge, we are not
aware of any reports in the literature describ-
ing the presence of central necrosis in ab-
dominal or pelvic lymph nodes in cervical
cancer when using helical CT or MR imag-
ing. This featureof heterogeneous en-
hancement of nodes with rim
enhancementmay prove useful in differen-
tiating benign from malignant disease be-
cause all nodes with necrosis on helical CT
and MR imaging were proven to be meta-
static in this study.

Helical CT

Attenuation of lymph nodes

.Attenuation
of lymph nodes on helical CT is important
because lymph nodes are recognized prima-
rily by the difference between their attenua-
tion and that of surrounding fat. Most lymph
nodes in this study were isodense to muscle
on unenhanced images, with no hyperdense
(to muscle) nodes noted.

Contrast enhancement.

Contrast-enhanced
helical CT has not, to our knowledge, been as-
sessed in the evaluation of pelvic lymph nodes
in cervical cancer. Patients are advanced
through the scanner at a continuous rate, thus
eliminating interscan delay and respiratory mis-
registration with helical CT. With bolus contrast
administration and rapid volumetric data acqui-
sition during peak vascular enhancement,
opacication of major pelvic arteries and veins
is excellent [20]. It would seem reasonable to
expect increased diagnostic condence in dif-
ferentiating vessels from lymph nodes. In the
pelvic cavity, there is a long interval between
the injection of contrast material and the time of
maximum enhancement of veins. Two-phase
helical CT was performed in this study to ad-
dress the difference in time of maximal en-
hancement between arteries and veins. No
signicant difference in the attenuation value of
the nodes in patients with benign and malignant
disease was noted on arterial and venous phase
scans. Likewise, the difference in the attenua-
tion value from arterial to venous phase was not
different in nodes from patients with benign and
from those with malignant disease. It may thus
be impossible to differentiate between true- and
false-positives by contrast enhancement or den-
sity alone. The additional cost of two-phase CT
does not seem justied by the lack of additional
advantage. Later (venous phase) imaging is
probably the most important for lymph node de-
tection to differentiate nodes from unopacied
veins [21] and therefore is probably preferred if
single-phase imaging is used in the pelvis.

MR Imaging

Signal intensity

.The MR characteristics
of enlarged nodes have been described as ho-
mogeneous on enhanced T1-weighted im-
ages. The presence of central necrosis, best
shown on fat-suppressed and unenhanced
T2-weighted sequences, has previously been
described as unhelpful in differentiating met-
astatic from nonmetastatic pelvic nodes in
cervical cancer [1]. Our ndings in this study
suggest that central necrosis may prove to be
Fig. 6.Bar chart displays distribution of peak rate of enhancement on MR imaging for nodes in patients with
histologic benign (white bars) and malignant (black bars) disease. Malignant nodes showed earlier peak rate of
enhancement compared with benign nodes, but difference was not signicant (p > 0.5).
D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
j
r
o
n
l
i
n
e
.
o
r
g

b
y

1
1
4
.
4
.
2
3
.
0

o
n

0
9
/
0
8
/
1
4

f
r
o
m

I
P

a
d
d
r
e
s
s

1
1
4
.
4
.
2
3
.
0
.

C
o
p
y
r
i
g
h
t

A
R
R
S
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
;

a
l
l

r
i
g
h
t
s

r
e
s
e
r
v
e
d


766

AJR:175, September 2000

Yang et al.

of value in the diagnosis of metastatic pelvic
nodes in cervical cancer. Despite a lower un-
enhanced signal intensity in malignant nodes
when compared with benign nodes, signi-
cant overlap in signal intensity for benign
and malignant nodes was observed. Thus, the
ability of MR imaging to characterize lymph
nodes on the basis of tissue characteristics as
established by signal intensity appears lim-
ited. It seems likely that increased size (gen-
erally >1 cm) and morphologic changes will
remain the primary parameters for identify-
ing abnormal nodes on MR imaging.

Contrast enhancement.

MR imaging in the
neck has shown heterogeneity of signal inten-
sity on T2-weighted and contrast-enhanced



T1-
weighted images to be a good indicator of
metastases in cervical nodes [18, 22, 23]. Heter-
ogeneity of signal intensity has also been noted
within a pelvic node in a patient with non-
Hodgkins lymphoma that correlated with cen-
tral necrosis on CT [24]. Other authors have
found it difcult to categorize pelvic lymph
nodes on T2-weighted images because of arti-
facts and poor denition of nodes [1]. Our nd-
ings conrm previous ndings that there is no
signicant difference in the degree of contrast
enhancement in nodes from patients with meta-
static and from those with nonmetastatic disease
[1]. The role of dynamic contrast-enhanced MR
imaging in the diagnosis of metastatic pelvic
lymph nodes appears to be limited. The rate of
enhancement, as well as time of peak enhance-
ment as determined by the slice number in the
dynamic sequence, showed no difference in
nodes from patients with metastatic and from
those with nonmetastatic disease.
In conclusion, size as determined by maximal
axial diameter and minimal axial diameter was a
good discriminant between benign and malig-
nant lymphadenopathy, but shape as determined
by the maximal axial diameterminimal axial di-
ameter ratio was not effective. Central necrosis
had a positive predictive value of 100% in the di-
agnosis of metastasis in pelvic lymph nodes in
this study and may be a valuable diagnostic fea-
ture when present on both helical CT and MR
imaging. In patients with metastatic nodes at pa-
thology, the densityenhancement pattern on CT
and the signal intensity on MR imaging did not
differ from measurements obtained from pa-
tients with benign nodes. The limitation of this
observation however is the issue of pathologic
proof of benign versus malignant nodes because
individual nodes were not specically sampled
on CT and MR imaging. Obtaining histology of
individual nodes identied on CT or MR imag-
ing will likely remain technically challenging,
although the advent of laparoscopic guidance for
pelvic lymph node biopsy may provide a tool by
which to sample suspicious nodes identied at
cross-sectional imaging.

References

1. Kim SH, Kim SC, Byung IC, Han MC. Uterine
cervical carcinoma: evaluation of pelvic lymph
node metastasis with MR imaging.

Radiology

1994

;190:807811
2. Kim SH, Choi BI, Han JK, et al. Preoperative
staging of uterine cervical carcinoma: compari-
son of CT and MRI in 99 patients.

J Comput As-
sist Tomogr



1993

;17:633640
3. Kim SH, Choi BI, Lee HP, et al. Uterine cervical
carcinoma: comparison of CT and MR ndings.

Radiology



1990;175:4551
4. Hricak H, Lacey CG, Sandles LG, Chang YCF,
Winkler ML, Stern JL. Invasive cervical carci-
noma: comparison of MR imaging and surgical
ndings. Radiology 1988;166:623631
5. Hricak H, Powell CB, Yu KK, et al. Invasive cer-
vical carcinoma: role of MR imaging in pretreat-
ment work-upcost minimization and diagnostic
efcacy analysis. Radiology 1996;198:403409
6. Roy C, Le Bras Y, Mangold L, et al. Small pelvic
lymph node metastases: evaluation with MR im-
aging. Clin Radiol 1997;52:437440
7. Yu KK, Hricak H, Subak LL, Zaloudek CJ, Pow-
ell CB. Preoperative staging of cervical carci-
noma: phased array coil fast spin-echo versus
body coil spin-echo T2-weighted MR imaging.
AJR 1998;171:707711
8. Fukuda H, Nakagawa T, Shibuya H. Metastases
to pelvic lymph nodes from carcinoma in the pel-
vic cavity: diagnosis using thin-section CT. Clin
Radiol 1999;54:237242
9. Fukuya T, Honda H, Hayashi T, et al. Lymph
node metastases: efcacy of detection with heli-
cal CT in patients with gastric cancer. Radiology
1995;197:705711
10. Steinkamp HJ, Hosten N, Richter C, Schedel H,
Felix R. Enlarged cervical lymph nodes at helical
CT. Radiology 1994;191:795798
11. Remy-Jardin M, Duyck P, Remy J, et al. Hilar
lymph nodes: identication with spiral CT and his-
tologic correlation. Radiology 1995;196:387394
12. American Joint Committee on Cancer. Manual for
staging of cancer, 3rd ed. Philadelphia: Lippincott,
1988:151153
13. Castellino RA, Marglin MI. Imaging of abdominal
and pelvic lymph nodes: lymphography or com-
puted tomography? Invest Radiol 1982;17:433443
14. Park JM, Charnsangavej C, Yoshimitsu K, Herron
DH, Robinson TJ, Wallace S. Pathways of nodal
metastasis from pelvic tumors: CT demonstra-
tion. RadioGraphics 1994;14:13091321
15. Vinnicombe SJ, Norman AR, Nicolson V, Hus-
band JE. Normal pelvic lymph nodes: evaluation
with CT after bipedal lymphangiography. Radiol-
ogy 1995;349355
16. Magnusson A. Size of normal retroperitoneal
lymph nodes. Acta Radiol Diagn 1983;24:315318
17. Hricak H, Yu KK. Radiology in invasive cervical
cancer. AJR 1996;167:11011108
18. Lien HH. MR imaging of invasive carcinoma of the
uterine cervix. Acta Radiol 1999;40:236245
19. Stark DD, Moss AA, Gamsu GT, Clark OH,
Gooding GAW, Webb WR. Magnetic resonance
imaging of the neck. II. Pathologic ndings. Ra-
diology 1984;150:455461
20. Urban BA, Fishman EK. Spiral CT of the female
pelvis: clinical applications. Abdom Imaging
1995;20:914
21. Teefey SA, Baron RL, Schulte SI, Shuman WP. Dif-
ferentiating pelvic veins and enlarged lymph nodes:
optimal CT techniques. Radiology 1990;175:683685
22. Som PM. Detection of metastasis in cervical
lymph nodes: CT and MR criteria and differential
diagnosis. AJR 1992;158:961969
23. Van den Brekel MWM, Castelijins JA, Stel HV, et
al. Detection and characterization of metastatic
cervical adenopathy by MR imaging: comparison
of different MR techniques. J Comput Assist To-
mogr 1990;14:581589
24. Lee JKT, Heiken JP, Ling D, et al. Magnetic reso-
nance imaging of abdominal and pelvic lymphad-
enopathy. Radiology 1984;153:181188
D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
j
r
o
n
l
i
n
e
.
o
r
g

b
y

1
1
4
.
4
.
2
3
.
0

o
n

0
9
/
0
8
/
1
4

f
r
o
m

I
P

a
d
d
r
e
s
s

1
1
4
.
4
.
2
3
.
0
.

C
o
p
y
r
i
g
h
t

A
R
R
S
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
;

a
l
l

r
i
g
h
t
s

r
e
s
e
r
v
e
d