Ruben
Kier,
Renal with
MD
#{149} Kenneth
Masses: Doppler
J. W. Taylor,
terms:
Angiomyolipoma,
91.3141
hemorrhage, 91.413 #{149} Kidney, infection, 91.212 #{149} Kidney neoptasms, US studies, 81.12984 #{149} Ultrasound (US), Doppler studies, 81.12984 Kidney,
cysts,
Radiology
91.31
1990;
Kidney,
#{149}
176:703-707
LTHOUGH
From the Department of Diagnostic Radiology, Yale University School of Medicine, 333 Cedar St, P0 Box 3333, New Haven, CT 06510. From the 1989 RSNA scientific assembly. Received January 10, 1990; revision requested March 8; final revision received May 10; accepted May 1 1 . Supported by American Cancer Society grant PDT 936 and National Institutes of Health grant ROl CA41729-01 Al. Address C
requests
RSNA,
1990
to R.K.
#{149} Ann
computed
(CT) and bowed improved renal
mass
have
addressed
L Feyock,
RT, RDMS
that
raise
such tions,
tomography
sonography detection
lesions
(1-3),
the
few
(4). Many imaging
concern
cyst
articles
of nerenal findings
for malignancy,
as septations, small or high attenuation
Although
have a!of small
specificity
nal mass detection cystic lesions have
calcificaat CT
puncture
and
(5).
opacifi-
cation was occasionally performed, most of these patients are followed up with serial imaging. Fine-needle aspiration of renal mass lesions unden sonographic or CT guidance has
been
disappointing,
with
ity of only 20%-62% thusiasm about the
renal nance
a sensitiv-
(2). Despite encapability to stage
neoplasms with magnetic reso(MR) imaging (6), MR imaging
has proved to be relatively limited in the detection and characterization of renal masses (7,8). Doppler ultrasound (US) has shown high sensitivity in the detection of tumor vasculanity within a variety of malignant lesions involving the liver, kidneys, adrenal gland, and
pancreas shown quencies carcinomas spectively nal mass established
(9,10). abnormal
Two studies have Doppler shift fre-
in 78%-83% of renal cell (1 1,12). This report proanalyzes indeterminate nelesions with use of criteria during one of those earli-
en retrospective
studies
MATERIALS
AND
During through nal mass
reprint
PhD
#{149} Isabel
M. Ramos,
MD
Characterization US’
The vascularity of indeterminate renal masses in 70 patients was investigated prospectively with duplex ultrasound. The peak-systolic Doppler shift frequency obtained from the renal mass was utilized to attempt distinction between benign and malignant lesions. With use of the criterion of a peak-systolic Doppler shift frequency of 2.5 kHz or greater as evidence of neovascularity, 26 of 37 malignant lesions demonstrated tumor signals (70% sensitivity). Thirty-one of 33 benign lesions lacked tumor signals (94% specificity). Both of the false-positive lesions were infections with inflammatory masses, with peak frequencies of 3.0 and 3.7 kHz. Tumor vascularity in most malignant renal mass lesions gives rise to abnormal, high-frequency, Doppler-shifted signals that can aid the differential diagnosis of renal masses. Index
MD,
basis
studies
of
tially with sonography (18 masses), or excretory masses). For 36 masses, nosis was subsequently either surgery the diagnosis
(42 masses), CT urography (10 a histologic diagmade by means
or biopsy. For was established
(a) definitive simple cyst
correlative at follow-up
phy, fat at CT [angiomyolipoma], mal parenchymal enhancement [ prominent lobation, columns
tion
of inflammatory lesions),
or (c) clinical
equipment
(Mark
vanced Wash)
was
ating ten
600
Technology used
in all
frequency was
Hz).
set
was
as low
Sample
volume
mass,
shift the
UltraMark
8; Ad-
The
wall
(50
or
volume
was
astolic frequencies were the Pourcelot index was
recorded, calculated.
the artery
US
and
ipsilateral the
corded.
was
or
size
METHODS
tumor tively
signals were as malignant probability
and In 24 main
Doppler
reTumor
frequency
shift frequencies (11). Renal masses
The
defined
to
Doppler and end-di-
with
was
a
the
contralateral
assessed
peak-systolic
Lesion
Once
reduced
the
renal
at 10
around
5 or 3 mm to better define shift frequency. Peak-systolic
cases,
fil100
manipulated of the mass.
detected
sample
inson-
The
was set initially was
was
Bothell,
studies. 3 MHz.
as possible
mm, and the cursor around the periphery
Doppler
or
Laboratories,
were
patients.
fol-
low-up for a minimum of 13 months. Commercially available duplex US
Doppler greater
105
a minienresolu-
or stability
lesions,
benign
signals
Doppler sonography of these masses was used prospectively to attempt their categonization as either benign or malignant, with use of criteria established in a prior retrospective series of 49 patients (1 1). Masses that were diagnosed confidently as benign (eg, simple cyst) on initial imaging studies were not referred for Dopplen sonognaphy.
or norat CT of Bertin,
and scar]), (b) serial imaging for mum of 5 months (eg, progressive largement of malignant lesions, of other
of
34 masses, by either
imaging (eg, CT or sonogra-
(1 1).
the period from October 1987 January 1990, indeterminate relesions were diagnosed on the
of imaging
Seventy renal masses with an established diagnosis constitute the study population. These lesions were imaged mi-
measured.
as peak-systolic
of 2.5 kHz demonstrating
diagnosed lesions.
or
prospec-
of malignancy
based
on
imaging findings alone was assessed retrospectively. The probability was considered high for solid or complex masses measuring
6 cm
or larger
and
low
for
masses 3 cm on smaller that were either (a) isoattenuated or isoechoic to normal renal parenchyma on the original imaging study remaining
or (b) predominately renal masses were
cystic. assigned
termediate probability. These retrospective assignments were then correlated with Doppler data and final diagnosis
The in-
to
703
KHz P
E A K
8
S
*
rO.O9
*
14
N U M B E
-
10
12
R
V
BENIGN
MALIGNANT
10
0
S
F
T
0
R
E N
Ii C
I:
F R
E Q U E
:
N
C V
0
**
MALIGNANT
LESION
DIAMETER
A
L M A
S S E S
#{176}
0
O11
(CM)
PEAK
1.
2.5-4
SYSTOLIC
DOPPLER
4.1-6
SHIFT
6.1-8
(KHz)
2.
Figures
1, 2.
relation
coefficient
(1) Correlation
of the
equaled
.09,
size
of malignant
indicating
renal
no correlation.
lesions
with
peak-systolic
(2) Distribution
Doppler
of peak-systolic
shift
frequency
frequency shifts
obtained for
benign
from and
the
lesion.
malignant
Cormasses.
Il
S T
0
0
L
0 G
*
$ARC
0
METS
5
MISC
0
WILM
x
V
ol
2.5
0
5
PEAK
Figure
3.
shift
DOPPLER
Malignant
according tnibuted
7.5
SYSTOLIC
renal
SHIFT
lesions
to 1KHz)
divided
to histologic diagnosis according to peak-systolic
and disDoppler
frequency.
TCCA transitional-cell SARC = sarcoma, METS metastases to the kidney, MISC miscellaneous malignancy, WILM = Wilms tumor, RCCA renal cell carcinoma.
carcinoma,
assess
the
in each The
relative
of those study
was
Investigation
utility
three
of Doppler
US
groups.
approved
Committee
by
the
of Yale
Human Univer-
sity, and verbal consent was obtained from the patient or a family member. The spatial peak/average intensity was reduced
with
the
equipment
attenuator
by
-9 dB to approximately 120 mW/cm2 with further tissue attenuation reducing this
to less
than
40
mW/cm2
at the
tumor
benign
Mean
systolic
frequency
artery
was .
the Pourcebot inratio of the peak-
significantly
Radiology
to the
masses
Figure
Thirty-seven renal masses were malignant and 33 masses were benign. The Doppler data are summanized in Table 1 . Peak-systolic frequency shift was significantly different between benign lesions (mean, 0.9 kHz) and malignant lesions (mean, 3.6 kHz). End-diastolic frequency shift was also significantly
although not. The
size
are
was
did
the peak-systolic (Fig 1).
RESULTS
different, dex was
and
data
main
different
renal
be-
malignant
tion
from
not
± 3.6 for maof malignant
correlate
with
frequency
2 demonstrates malignant
only
3.4 cm ± 2.0 for be-
of peak-systolic
among
lesions,
available
nign lesions and 6.6 cm lignant lesions. The size
renal
site.
704
tween although 24 cases.
shift
the
distnibu-
formed
masses. Whereas the majority of malignant lesions (26 of 37) had a peaksystolic frequency shift of 2.5 kHz or greater (70% sensitivity), only two of the 33 benign lesions had a shift in this range (94% specificity). Figure 3
separates shift
the
peak-systolic
of malignant
to their histologic teen of 24 renal had
peak-systolic
lesions
cell
frequency according
diagnosis. carcinomas
Doppler
kidney
known where ment
benign
Eigh(75%)
shifts
of
4-7).
All four peak of 2.5 kHz or
greater. All three cases of transitional cell carcinoma demonstrated low frequency shifts. The misceblaneous category refers to large mass lesions in the
frequency
versus
2.5 kHz or greater (Figs Wilms tumors demonstrated systolic frequency shifts
in patients
who
primary malignancy but in whom histologic of the
kidney
was
to determine
according
per-
the
tumor
ne-
or a
lesions
to their
a
elseassess-
whether
nal mass was a primary metastasis to the kidney. The 33 benign renal
classified
not
had
are
diagnos-
tic categories in Figure 8. Only two benign renal masses demonstrated peak-systolic frequency shifts over 2.5 kHz. These exceptions occurred in two of eight cases of renal infection that demonstrated frequency shifts of 3.0 and 3.7 kHz. Both pa-
September
1990
a.
b.
Figure
4.
Renal
cell
carcinoma
C.
in a patient
with
chronic
renal
failure
undergoing
peritoneal
dialysis.
(a) CT scan
demonstrates
a complex
cystic mass in the left kidney (arrow). Adjacent images demonstrated simple cysts in both atrophic native kidneys. Free fluid in the left pencolic gutter is from penitoneal dialysis. (b) Sonogram of the left kidney demonstrates a complex cystic mass. The Doppler cursor was manipulated around the periphery of this mass to obtain Doppler signals. (c) Doppler spectrum of the left renal mass demonstrates peak-systolic frequency of almost 6 kHz, consistent with malignancy.
b.
a. Figure
5.
Renal
cell
carcinoma
C.
manifesting as hematuria. is slightly hyperattenuated
the left kidney (arrow) that mass to be slightly hypoechoic. The Doppler peak-systolic frequency of 4 kHz, consistent
(a) CT scan obtained with respect to normal
cursor is at the periphery with malignancy.
83%
assigned an intermediate probability, with malignancy diagnosed in 12 (28%). Table 2 assesses the utility of
cy shifts
of 2.5
hypovasculan Volume
kHz
or greater.
or avasculan 176
#{149} Number
3
Of
masses
11
carcinoma,
(c) Doppler
(five
Seventeen masses (14 malignant, three benign) were assessed with angiography. Of six masses (all renal cell carcinoma) that were hypervascular or vascular at angiography, five demonstrated peak-systolic frequen-
cell
of the mass.
tients demonstrated clinical evidence of infection. Among patients in whom the main renal artery was sampled, the renal mass frequency shift exceeded that of the artery for 1 1 of 12 malignant lesions and two of 12 benign lesions (92% sensitivity, 83% specificity, 87% accuracy). In this same group, a peaksystolic frequency cutoff of 2.5 kHz was 75% sensitive, 92% specific, and accurate.
renal
without intravenous renal parenchyma.
two
transi-
tional cell carcinoma, one sarcoma, and three benign masses), nine had Doppler shifts under 2.5 kHz.
On
the
basis
of imaging
masses were assigned bility of malignancy, cy proved in 17 (85%).
were
assigned
malignancy
(14%).
Doppler
a low discovered
The
remaining
US in detecting
alone,
probability,
with one
43 masses
were
malignancy
in masses with these various possibilities for malignancy. the intermediate probability ry, Doppler detection of tumor nals was 58% sensitive (11 of
contrast Sonognam
spectrum
pretest Among categosig19) and
material of the
shows a 2-cm mass left kidney shows
of the left renal
96% specific renal
mass
(23 of 24) for
in the
demonstrates
malignant
masses.
DISCUSSION
20
a high probawith malignanSeven masses in only
(b)
This prospective series mass lesions demonstrates frequency Doppler-shifted
are renal nign port
encountered
(11,12). prior from (12),
in most
mass lesions renal masses. the conclusions
Our
of 70 renal that highsignals
malignant
yet are rare in beThese results supof earlier reports
current
study
and
our
retrospective study (1 1 ) differ that of Kuijpers and Jaspers based on the cutoff value we
recommended.
Whereas
seven
of
nine renal cell carcinomas had Dopplen shifts of 4 kHz or higher in their series (78%), only 1 1 of 24 carcinomas
Radiology
#{149} 705
a.
b.
Figure
Renal
6.
Doppler consistent
cursor with
cell carcinoma is on the malignancy.
inner
manifesting
C.
as microhematunia.
margin of the mass. (c) Left renal angiogram
(b)
(a) Sonogram
demonstrates
Doppler interrogation shows a hypervascular
of the mass mass (small
a 2-cm demonstrates arrows) with
isoechoic
left renal
mass
(arrows).
peak-systolic frequency an early draining vein
The
of 8 kHz, (large arrow).
in our current series had frequency shifts in this range (46%). Even if our two cases of miscellaneous malignancy were renal cell carcinomas, the sensitivity of using a cutoff level of 4
kHz would differences
increase between
may reflect the small the study of Kuijper
achieve
reasonably
to only 50%. these studies
The
sample size and Jaspers.
in To
high
sensitivity
for malignancy, we recommend a cutoff value of 2.5 kHz. This criterion provides high specificity (94%), with false-positive diagnoses limited to two cases of renal infection. Since renal infection is often clinically apparent, false-positive diagnoses based on Doppler investigation should be rare. Although comparison of renal mass Doppler signals to the main renab artery Doppler signals may be as useful as peak-systolic frequency for detection of malignancy (11), this approach may be less practical. First, an accurate sampling of the main renal
artery
was
occasionally
difficult
a.
b.
Figure
7. Renal gram demonstrates of the renal mass tinuous
waveforms
respiration,
carcinoma in a patient with benign prostatic hypertrophy. (a) Sonoa 4-cm hypoechoic right renal mass (arrows). (b) Doppler interrogation shows peak-systolic frequency of 3 kHz, consistent with malignancy. Conwere
such
difficult
as in this
to obtain
--
some
patients
who
were
unable
to suspend
HEMATOMA +
*
ANOIOMI’5 LOBATION
to
obtain, which may relate partially to operator dependence. Second, sampling the main renal artery increased study time without significantly improving accuracy. Doppler shift frequencies reflect the velocity of blood flow within a mass. Recent studies have demonstrated the histologic and angiographic presence of arteniovenous
in
case.
-
D I
I
cell
0
COLUMN
S 08
0 PEAK
2.5 SYSTOLIC
5 DOPPLER
7.5 SHIFT
Figure 8. Benign renal lesions cording to the clinical diagnosis uted according to peak-systolic shift frequency. ANGIOMYO poma.
(KHz)
divided acand distnibDoppler angiomyoli-
and hypervasculanity in masses with high Doppler frequency (9,10,12). Kuijpers and Jaspers noted that all seven renal cancers with Doppler shifts of 4 kHz on more were hypervascular or vascular at angiography, whereas two masses with
cular at angiography (12). Our study showed a similar but slightly weaker correlation of Doppler shifts with angiographic vasculanity. Correlation may not be perfect since Doppler fre-
whereas angiognaphic vasculanity is determined by the total volume of flow. We are currently performing a retrospective study to ascertain which specific angiognaphic features
shifts
quencies
may
shunts
706
of
2 kHz
#{149} Radiology
on
less
were
hypovas-
reflect
velocity
of
flow
correlate
most
closely
with
September
1990
high-frequency We conclude characterization sions.
We
tigation sion phy,
References
Doppler signals. that Doppler US aids of renal mass be-
recommend
whenever
Doppler
a renal
is encountered during since this additional
1.
exclude
malignancy.
Acknowledgment: knowledge statistical
Volume
Robert
The Lange,
sonogna-
analysis.
176
#{149} Number
3
gratefully for providing
Horii
2.
carcinomas:
cell dilemma.
BN.
discovery
Radiology
Bree
and
RL, Pollack
of
HM,
et al.
resolving
a
9.
1988;
excretory
L, et
imaging
features.
AJR
10.
6.
Fein AB, Lee JKT, Baife DM, et a!. Diagnosis and staging of renal cell carcinoma: a comparison of MR imaging and CT. AIR
7.
Quint LE, Glazer GM, Chenevert In vivo and in vitro MR imaging
1987;
11.
ap-
158: 1-10.
ac-
correlation and Radiology
1988; 169:359-362. Manotti M. Hnicak H, Fnitzsche P. Crooks LE, Hedgcock MW, Tanagho EA. Complex and simple renal cysts: comparative evaluation with MR imaging. Radiology 1987; 162:679-684. Taylor KJW, Ramos I, Carter D, Morse 55, Snowen D, Fortune K. Correlation of Doppler US tumor signals with neovasculan morphologic features. Radiology 1988; Taylor
KJW,
masses: Doppler Ramos Snowen
Ramos I, Morse 55, Fortune DO, Taylor CR. Focal liven differential diagnosis with pulsed US. Radiology 1987; 164:643-647. IM, Taylor KJW, Kier R, Burns PN, DP, Carter D. Tumor vascular
signals
in
KL, Hammers
1989;
MA. The current radiological to renal cysts. Radiology 1986;
proach
histopathologic sequence optimization.
166:57-62.
urogra-
phy/linean tomography, US, and CT. Radiology 1988; 169:363-365. Levine E, Huntrakoon M, Wetzel LH. Small renal neoplasms: clinical, patholog153:69-73. Bosniak
8.
1989;
Radiology
specificities
Reand
166:637-641. Warshauen DM, McCarthy SM, Street al. Detection of renal masses: sensitiv-
ic, and 5.
detection.
renal
AJ, Hul-
Raghavendra earlier
Small
ities
4.
SC,
Megibow
170:699-703. Amendola MA, diagnostic
3.
MA,
carcinoma:
increased
be-
U authors PhD,
DH,
nal-cell
investigation will add only a few minutes to the patient’s study. Furthenmore, duplex US should be considered to enable further characterization of equivocal mass lesions discovered at intravenous urognaphy, CT, or angiognaphy. In the absence of clinical evidence of infection, a Doppler shift frequency of 2.5 kHz or greater is a reliable indicator of malignancy. However, the absence of a high-frequency Doppler shift does not
SJ, Bosniak
nick
inves-
mass
tumors: pulse
Smith
12.
renal
masses:
detection
with
Doppler US. Radiology 1988; 168:633-637. Kuijpens D, Jaspens R. Renal masses: differential diagnosis with pulsed Doppler US. Radiology 1989; 270:59-60.
148:749-753.
TL, et al. of renal
Radiology
#{149} 707