Richard C. Semelka, MD •¿ Hedvig Hricak, MD •¿ Susan K. Stevens, MD •¿ Robert Finegold, MD •¿ Ernesto Tomei, MD •¿ Peter R. Carroll, MD
Combined Gadolinium-enhanced and Fat-saturation MR Imaging of Renal Masses' Combined gadopentetate dimeglu mine enhancement and fat-satura tion (FS) spin-echo (SE) magnetic resonance (MR) imaging for the de tection and characterization of renal masses was evaluated in 43 patients with a total of 71 lesions (28 solid masses and 43 cysts). SE MR se quences compared were the follow ing: short repetition time (TR)/echo time (TE), conventional SE, short TR/TE FS SE, long TRITE conven tional SE, gadolinium-enhanced short TRITE conventional SE, and gadolinium-enhanced short TRITE FS SE techniques. MR findings were compared with findings of contrast enhanced computed tomography (CT) and with pathologic findings in all patients. The sensitivities for detection of renal masses with gado linium-enhanced FS (71 of 71 le sions) and with gadolinium-en hanced short TRITE conventional (65 of 71 lesions) SE sequences were significantly (P < .01) greater than with any unenhanced (short TRITE conventional [40 of 71 lesions], short TR/TE FS [44 of 71 lesions], or long TR/TE [39 of 71 lesions]) SE se quence. Lesion characterization was also best with the gadolinium-en hanced FS SE sequence (65 of 71 le sions correctly classified). When combined pre- and postcontrast short TRITE FS SE images were ana lyzed with both qualitative (visual) and quantitative (region-of-interest measurements) assessment, lesion characterization improved even fur them (70 of 71 lesions were correctly characterized). All lesions detected with CT were visualized with the gadolinium-enhanced FS SE MR se quence, which in addition depicted seven cysts and two small renal cell carcinomas. In summary, the use of gadopentetate dimeglumine, espe cially when combined with the FS technique, was superior to unen hanced MR imaging for detection and characterization of renal lesions.
P
RESENTLY, magnetic
have been used to decrease chemical
resonance
(MR) imaging for the evaluation of renal masses has been reserved for selected cases of renal cell carcinoma in which either tumor thrombus in the renal vein and/or inferior vena cava or invasion of adjacent organs is suspected (1-8) but not documented at computed tomography (CT). For the identification
of renal lesions,
es
pecially when tumor size is smaller than 3 cm, non-contrast-enhanced
MR images—because spatial
and contrast
of decreased resolution
and
increased artifacts—are inferior when compared with iodine contrast material-enhanced CT scans (1). Gadopentetate dimeglumine is an intravenous panamagnetic contrast agent distributed within the intravas culan and extracellulan space and is excreted in unchanged form by the kidneys system,
(9). In the central gadolinium-enhanced
nervous MR
imaging has been shown to improve lesion detectability as compared with unenhanced MR imaging or contrast enhanced CT (10-12). Previous clini cal reports have shown the potential value of contrast-enhanced MR imag ing in the evaluation
shift artifacts
for structures
surround
ed by fat (eg, the optic nerve) (15) and to decrease phase artifacts in the upper abdomen by reducing the sig nal intensity of the respiration-in
duced motion of subcutaneous fat (16). The combination of gadopente tate dimeglumine enhancement with fat-saturation technique has been useful for detecting various patho logic conditions in the head, breast, and musculoskeletal system by de creasing the high signal intensity of fat, thereby increasing the conspicu ity of contrast-enhanced abnommali ties (17). To our knowledge, theme are no reports in the literature on the use
of combined fat-saturation and gado linium-enhanced MR imaging in the abdomen. The purpose of this study was to evaluate the usefulness of gadopen tetate dimeglumine enhancement alone and in combination with fat saturation
MR imaging
for the assess
ment of renal masses. MATERIALS
AND
METHODS
of renal masses
(13,14). Fat-saturation
Patients
techniques
The study population consisted of 43 consecutive patients who, based on the Index terms: Gadolinium . Kidney, cysts, 81.312 •¿ Kidney neoplasms, diagnosis, 81.324 •¿ Kidney neoplasms, MR studies, 81.1214 •¿ Kid ney neoplasms, staging, 81.324 •¿ Magnetic res onance (MR), pulse sequences
abnormalities identified at CT examina tions, were considered for surgery or bi opsy. CT findings indicated the lesion of
primary concern to be either a solid mass
(n
30) or complex renal cyst (n
13).
Before surgery or biopsy, all 43 patients Radiology
underwent MR imaging. The patients, 30 men and 13 women, ranged in age from 31 to 78 years (mean, 61 years). Surgical
1991; 178:803—809
(n
40) on biopsy (n
3) confirmation
was obtained within 2 weeks after the 1 From
the
Departments
of
Radiology
(R.C.S.,
H.H., S.K.S., E.T.) and Urology (H.H., P.R.C.), University of California School of Medicine, Box 0628, San Francisco, CA 94143-0628; and the Department of Radiology, Letterman Army Medical Center, San Francisco (R.F.). Received February 26, 1990; revision requested April 10; final revision received October 3; accepted Oc toben 9. Address reprint requests to H.H. ‘¿c RSNA, 1991
MR imaging study. Surgical procedures included exploration with lesion enucle ation (n = 7), and segmental (n = 9),
Abbreviations: ROI region of interest, SE = spin echo, SD/N = signal-difference-to-noise ratio, TE = echo time, TR repetition time.
803
@
Figure 1. Axial images of simple renal cyst (incidental finding in a patient with renal cell carcinoma in the inferior pole of the left kidney) obtained at the same anatomic level:
(a) contrast-enhanced
CT scan, (b) short TR/
TE (700/20) SE image, (c) short TR/TE (700/ 20) gadolinium-enhanced SE image, and (d) gadolinium-enhanced fat-saturation (700/ 20) image. Two renal cysts in the left kidney are appreciated on the SE and fat-saturation sequences enhanced with gadopentetate di meglumine. The unenhanced SE sequence and the CT scan do not depict these cysts, which are less than 4 mm in diameter. The two small cysts have sharp outlines on the fat-saturation image (arrows in d), and al though identified with gadopentetate dime glumine-enhanced conventional SE image
(c), they are not as sharp in detail.
simple
(n = 11), or radical
(n
a.
b.
13)
nephrectomy. Contrast-enhanced CT scans (36 studies on GE 9800 [GE Medical Systems, Mil
waukee], four on GE 8800, two on Sie mens tems,
Somatom [Siemens Medical Sys Iselin, NJ], and one on Picker 1200
[Picker International, Highland Heights, Ohio] CT scanners) were obtained in all patients imaging
within study.
2 weeks before CT examinations
the MR were
performed at different institutions and were not performed with a uniform pro
d.
C.
tocol. All CT studies, however, were per formed with a bolus contrast material technique, with 1-cm contiguous sections
obtained through
the kidneys. Thirty
three studies were performed before and after injection of contrast material, and 10 studies were performed only after the in jection of contrast material.
Imaging
Technique
MR imaging was performed
on a 1.5-T
cryogenic imagen (Signa; GE Medical Sys tems). All patients gave informed consent
with approval of the Human Research Committee of the University of California at San Francisco, prior to an MR imaging study. Short repetition time (TR) msec/ echo time (TE) msec (300—600/ 15—30)im aging was performed in the axial plane
with a matrix size of 192 X 256 and two excitations. Section thickness was 8—10 mm, with a 1.0-2.5-mm intersection gap. Inferior and superior saturation pulses
were applied. These parameters
were
maintained as similar as possible between regular spin-echo (SE) and fat-saturation sequences. Slight variations were neces sary to ensure complete imaging of both kidneys, since the fat-saturation tech nique results in the loss of approximately 15% of the sections obtained with regular SE sequences. Variations between imag ing parameters for regular SE and fat-sat uration sequences never exceeded 20%. Pnecontrast long TR/TE (2,000-2,500/2070) images were obtained in all patients.
A bolus of gadopentetate dimeglumine was injected by hand at a dose of 0.1
804 •¿ Radiology
mmol/kg, with approximate injection time of 5—10seconds. Imaging parameters for postcontrast sequences were identical
to those of the precontrast sequence, with image acquisition commencing approxi mately 1 minute after injection. Typical imaging time for a study was 32 minutes, for a total study time of 60—75minutes. Fat saturation was achieved by applica tion of a low-amplitude, long-duration si nusoidal radio-frequency pulse at the fre quency of lipid followed by a section-se lect gradient pulse to disperse the fat
signal. The subsequent obtained
with
reception
SE sequence was centered
on the
frequency
of water.
The sequence
used
was the standard GE Medical Systems fat saturation
pulse
sequence.
Image Analysis Qualitative—The various individual MR imaging sequences and CT scans were randomly chosen by one of the au thors (E.T.) and were analyzed in confer ence by two radiologists (R.C.S., H.H.) without knowledge of the pathologic re suits. For each set of images, CT scan and/ or MR acquisition, the number and size of lesions were determined to compare the
March 1991
stage IV, one patient. The remaining 43 of 71 lesions were cystic, and all were benign. Cystic lesions included two calcified
(rim calcification)
cysts,
five hemorrhagic cysts (with septi and/or blood clots), three previously infected cysts, and 33 cysts for which no particular pathologic description was recorded. This was a retrospective
a.
b.
Figure 2. Axial images of 1-cm renal cell carcinoma in a patient with bilateral renal cell car cinoma. The cancer on the left side (not shown) measured many reason for image evaluation. (a) Contrast-enhanced hanced fat-saturation (600/ 15) MR image were obtained formation provided by the modalities is equivalent, with carcinoma in the anterolateral aspect of the night kidney
4 cm in diameter and was the pri CT scan and (b) gadolinium-en at the same anatomic level. The in demonstration of a 1-cm renal cell with central enhancement (arrow).
study,
and
detailed pathologic
correlation
not available,
than a review
other
was of
official pathology department me ports. The reports described the le sions as either malignant—in which case the detailed description of the tumor type and extent was avail able—or as benign cyst. In most cases, the description
included lesions.
of benign
cyst
the number and size of the Only the findings
of large
noise levels before and after injection of contrast material resulted in some cystic lesions displaying either a slight increase on decrease in signal intensity, for the cal
blood clots, calcaneous plates, thick
(ROI) quantitative results were used. Le
culation of percentage
sions were characterized as either cysts or solid masses by means of the following
hancement for renal cysts, a second value was determined in which the standard deviation of noise of the postcontrast im
tween “¿complex― cyst at CT and pathologic findings could not be made.
sensitivity
of each individual
pulse
quence for lesion detection. For lesion characterization, tative
assessment
combinations short
se
both quali
and region
of interest
of MR pulse sequences: (a)
TR/TE
and long TR/TE
SE, (b) short
TR/TE SE before and after injection of ga dopentetate dimeglumine, and (c) short TR/TE fat saturation before and after in
ages was normalized
of contrast en
to the standard devi
jection of gadopentetate dimeglumine. With qualitative assessment based on pre viously published criteria, the lack of a
ation of noise of the precontrast images in the following manner: percentage of con trast enhancement of renal tissue X (pre contrast SD noise —¿ postcontnast SD noise), where SD is standard deviation. Normalization of the standard deviation
discernible
of noise was not necessary for solid
wall, sharp delineation
surrounding
parenchyma,
contrast enhancement cystic lesions
from
and lack of
lesions,
were indicative of
(1—4,18,19). Lesion
enhance
ment after injection of gadopentetate di meglumine was indicative of solid masses (14). Quantitative measurements of the signal-difference-to-noise ratio (SD/N) of the renal lesions to renal panenchyma and percentage of contrast enhancement of renal lesions compared with that of pre contrast
images
obtained
with
the same
MR technique (see below) were used to further characterize all lesions. Qualita tive and quantitative results were then combined to assign the lesion to the cys tic or solid category, since the classifica tion of masses as indeterminate was not
permitted
for the data analysis.
since the solid masses
always
enhanced to a moderate degree, which was less than the enhancement of cortex. The standard percentage of contrast en hancement adequately demonstrated this. The image display adds 1,024 to pixel sig nal intensity, and therefore this number
was subtracted
from all ROI detenmina
panenchyma
RESULTS
and (b) the percentage
chyma,
renal
masses,
and renal
pared with that of precontrast tamed Noise
Pathologic
of renal panen cysts com
images ob
with the same MR technique. for both cystic and solid lesions
was measured by a large rectangular ROI anterior to the abdomen, and the stan dand deviation of noise was used for cal culations contrast
the following:
(postcontrast
Volume 178 •¿ Number 3
masses in 30 patients and as complex cysts in 13 patients. After the images
were reviewed and CT measurements obtained when needed, 26 lesions were considered solid and 35 lesions were considered cystic (17 lesions were considered complex-indetemmi nate and 18 were simple cysts). Three additional lesions were miscla8sified as solid (Table 1).
Seventy-one
renal lesions in 43 pa
MR Imaging
Findings
No patients were excluded from the study on the basis of poor-quality
MR images. The gadolinium-en hanced fat-saturation sequence
dem
onstrated more lesions than any oth en sequence (71 of 71 lesions) (Fig 1).
The sensitivity for detection of solid renal masses and renal cysts for all MR sequences and CT is shown in Table 1. In 64 of 71 lesions, contrast enhanced CT and gadolinium-en hanced fat-saturation imaging pro
tients were present at surgery or con firmed with biopsy. Of the 71 le sions, 28 were malignant, including
vided comparable information (Fig 2). Gadolinium-enhanced fat-satuma tion imaging depicted seven more
24 renal cell carcinomas,
cysts
tional
cell carcinomas,
two transi and two pheo
by SI
tients, the following stages were found: stage I, seven patients; stage
SI —¿ precon
trast SI) + pnecontnast SI X 100, where is signal intensity. Because differing
Findings
chromocytomas. Among the 24 renal cell carcinomas present in 18 pa
of SD/N. The percentage of enhancement was determined
The primary lesions of concern were initially classified as solid
Statistical analysis.—All quantitative measurements were expressed as mean values plus or minus the standard devi anion. Statistical significance was deter mined with the Scheffe test for compari son by multivaniate analysis. The true number of lesions, lesion characteniza tion, and tumor extent were determined from surgical and histologic findings.
expressed as (a) SD/N of renal lesions to renal
CT Findings
tions.
Quantitative—Signal-intensity mea sunements were obtained in openator-de fined ROIs. The quantitative data were
of contrast enhancement
septa, on puttylike contents indicat ing infections were described sepa rately. Therefore, correlation be
II, six patients; stage III, four patients;
and
two
more
solid
lesions
(Fig
3) than did contrast-enhanced CT. All cysts detected with the gadolini
um-enhanced
fat-saturation
tech
nique that were not detected with the contrast-enhanced short TR/TE SE sequence measured less than 6
Radiology •¿ 805
3.
4.
Figures 3, 4. Synchronous renal cell carcinomas. (3) Gadolinium-enhanced fat-saturation (600/ 15) axial MR image shows a 3.5-cm renal cell carcinoma medially (solid arrow) and a second 1.8-cm renal cell carcinoma (open arrow) laterally in the solitary night kidney. This second renal cell carcinoma was not demonstrated with any other MR imaging sequence, nor was it identified prospectively on the CT scan. (4) Gadolinium-enhanced fat-saturation (600/15) axial image shows central necrosis (a) in a renal cell carcinoma.
Figure 5. Tumorthrombus.Conventional SE (600/15) axial image with gadopentetate dimeglumine enhancement shows enhance ment of tumor thrombus, which extends into the inferior vena cava (arrow). A large left renal cell carcinoma is present (arrow head).
mm in diameter, and those not de tected with contrast-enhanced CT measured less than 5 mm. Both of the latter techniques inconsistently iden tified 4-mm cysts. All solid masses detected with the gadolinium-en hanced fat-saturation technique that were not detected with contrast-en hanced short TRITE SE on contrast enhanced CT sequences measured less than 2 cm. The minimum size of lesions detected with the various MR sequences and CT is displayed in Ta bie 2. The conspicuity of lesions smaller than 2 cm in diameter was depicted best on gadolinium-en hanced fat-saturation images. Figure 1 illustrates the ability of gadolini um-enhanced MR images to depict small cysts that, in this example, measured less than 4 mm in diameter and were not detected with unen hanced MR imaging sequences or with contrast-enhanced CT. For lesion characterization, both qualitative assessment and ROI quan titative results were used. The combi nation of pre- and post-gadolinium enhanced short TR/TE fat-saturation images enabled connect chamactemiza tion of more lesions than combined short TR and long TR unenhanced SE 806 •¿ Radiology
a.
b.
Figure 6. Both tumor thrombus and bland thrombus in the inferior vena cava. Gadolinium enhanced fat-saturation (700/20) images (a) at the level of renal hilum and (b) a more infeni or section. Enhancing tumor thrombus (arrows in a) in right renal vein extends into the in fenior vena cava. Bland thrombus (arrow in b) does not show enhancement, but marked en hancement of the wall of the inferior vena cava and penicaval tissue is present.
and pme- and post-gadolinium-en hanced short TR/TE SE images (Table 3). The presence of a sharp interface and lack of contrast enhancement of renal
cysts
was
best
demonstrated
on
gadolinium-enhanced fat-saturation images. For comparison, Figure 3 shows an intrapamenchymal tumor that lacks a sharp interface with renal pamenchyma on the gadolinium-en hanced fat-saturation image. Two le sions with calcified walls on gadolin ium-enhanced fat-saturation images were correctly classified as unen hanced cysts; these were interpreted as solid with contrast-enhanced CT. Another cyst was misclassified as sol id with both CT and the gadolinium enhanced fat-saturation technique. The two pathologically proved tman sitional
cell carcinomas
and
two
pheochnomocytomas were correctly classified as solid lesions; however, all four lesions were incorrectly diag nosed as renal pamenchymal lesions with both MR imaging and CT.
At visual assessment, five of the hemorrhagic and previously infected lesions were misdiagnosed after in jection of gadopentetate dimeglu mine. Those lesions were of high sig nal intensity on the unenhanced short
TR/TE
sequence
and
therefore
were misdiagnosed as enhancing af ten injection of gadopentetate dime glumine. Quantitative measurements were essential for the chamactemiza tion of hemorrhagic lesions. The SD/ N data for all MR sequences used are listed in Table 4. The SD/N values for cysts increased significantly (P < .01) with contrast enhancement for both
fat-saturation
and
regular
SE se
quences, while the SD/N values for solid masses increased significantly (P < .01) for fat-saturation sequences and
moderately
(P < .05) for regular
SE sequences. Table 5 shows the percentage contrast
enhancement
of normal
of me
nal pamenchyma, solid renal masses, and renal cysts with the conventional March 1991
Table 3 Lesion Characterization for Each MR Sequence Combination and CT with Qualitative and Quantitative Evaluation PostcontrastShort
Pre- and
PostcontrastFat-saturationLongTR/TE and TR/TE,PreUnenhanced QuantitativeQualitative ImagesShortTR/TEShortTR/TEQualitative
Lesion
Quantitative
CT
Renalcyst
26/2638/4159/6564/6565/7170/7161/64 Solidrenalmass 18/1934/40 25/2539/40 25/2537/43 28/2842/43 28/2835/38 Total correctly characterized20/22 Note—For qualitative evaluation, characterization of cystic lesions was based on sharp delineation from surrounding parenchyma and lack of contrast enhancement; characterization of solid masses was based on contrast enhancement. For quantitative evaluation, characterization was based on qualitative criteria plus ROI results. Numbers are expressed as number of lesions correctly characterized/number of lesions detected.
Table 4 Lesion-Renal Cortex SD/N for Short TRITE andMR MR Imaging Sequences after Administration of Gadopentetate Dimegluminebefore SequencePnecontrastPostcontrastLesion
SaturationRenal
Precontrast
Postcontrast
cyst Solid renal mass
—¿4.3 ±7.7 —¿2.4 ±3.7
—¿22.7 ±14.7* —¿7.4 ±6.6t
Fat SaturationFat —¿9.0 ±8.0 ±20.5* —¿2.1 ±5.5—39.3 —¿17.2 ±17.1*
crosis (Fig 4), which was not cleanly demonstrated on unenhanced MR images (n = 5); (b) enhancement of renal cell carcinomas that were poor ly enhanced on CT scans (n 3); (c) enhancement of tumor thrombus in the renal vein (Fig 5) (n = 4); (d) no enhancement of blood thrombus in the inferior vena cava (n = 1) (Fig 6); and (e) enhancement of lymphade nopathy
(n
3).
With regard to staging accuracy for renal cell carcinoma, 13 were correct ly staged with CT, and in 14 of 18 pa tients, renal cell carcinoma was con mectly staged with both gadolinium enhanced sequences. If stages I and II are grouped together, two staging en mors were made with CT and one staging error was made with MR im aging. In cases in which both MR im aging and CT resulted in misstaging, the error was due to the presence of enlarged nodes that were not the me suit of metastatic disease. In the sec ond case that was misstaged with CT, the error was due to the presence of renal vein thrombosis that was not detected with CT but was detected with MR imaging.
Note—Numbers are expressed as SD/N ±standard deviation. * Statistically
significant
(P