DOI:10.2214/AJR.04.1545
AJR 2006; 186:1639-1650
© American Roentgen Ray Society
Comparison of 16-MDCT and MRI for Characterization of Kidney Lesions
Ambros J. Beer,
Martin Dobritz,
Niko Zantl,
Gregor Weirich,
Jens Stollfuss and
Ernst J. Rummeny
Department of Radiology, Technische Universitaet Munichen, Ismaninger
Strasse 21, Munich, Germany, 81675.
Received January 25, 2005;
accepted after revision August 9, 2005.
Address correspondence to A. J. Beer
(beer{at}roe.med.tum.de).
Abstract
OBJECTIVE. The objective of our study was to compare the diagnostic
performance of 16-MDCT with that of MRI in the characterization of kidney
lesions.
SUBJECTS AND METHODS. Twenty-eight patients with kidney lesions
detected with sonography and requiring further evaluation were examined. MDCT
was performed in the unenhanced, arterial, and portal venous phases. MRI was
performed at 1.5 T with T2- and T1-weighted and dynamic gadolinium-enhanced
sequences. Consensus reading was done by two radiologists. Image quality was
rated on a four-point scale. Classification of lesions as surgical or
nonsurgical was done with five levels of confidence, and it was required that
a definite diagnosis be assigned to each lesion. The 1997 TNM classification
was used for staging. Statistical analysis was done by receiver operating
characteristic analysis or paired Student's t test. Histologic or
follow-up findings at least 12 months after the primary diagnosis served as
the standard of reference.
RESULTS. The image quality of MDCT (mean grade, 2.79 on a 0-3 scale)
was superior to that of MRI (1.93; p < 0.01). The area under the
curve for differentiating surgical from nonsurgical lesions was 0.979 for MDCT
and 0.957 for MRI with resulting sensitivity and specificity values of 92.3%
and 96.3% for MDCT and 92.3% and 91.3% for MRI. Sensitivity and specificity
for definite classification of the lesions were 93.8% and 68.4% for MDCT and
93.8% and 71.4% for MRI.
CONCLUSION. Both MDCT and MRI are excellent for differentiating
surgical from nonsurgical kidney lesions. Both methods have low specificity
for the differentiation of benign from malignant lesions.
Keywords: CT technique • genitourinary tract imaging • kidney • MDCT • MRI
Introduction
The widespread use of sonography, CT, and MRI in the field of
abdominal imaging has led to an increase in the incidental detection of renal
lesions, which require further diagnostic evaluation
[1]. This evaluation is most
often done with CT or MRI, both of which are well-established techniques in
the field of renal imaging
[1-5].
Although CT is the preferred technique because of availability, lower cost,
and spatial resolution, it is still unclear which method is best suited for
renal imaging with state-of-the-art techniques. There have been few
comparisons of MRI and CT in the same patient, and only single-detector or
four-slice CT was used in those studies
[1-4].
Furthermore, most reports of renal imaging with MRI and CT emphasize accuracy
in TNM staging of malignant renal lesions
[6-9].
An issue of equal importance, however, is the capability of MRI and CT to
provide enough information to allow differentiation of surgical from
nonsurgical kidney lesions incidentally de tected with another imaging
technique, such as sonography or, less often, excretory urography. Surgical
lesions in this context are lesions in which malignancy cannot be excluded on
the basis of imaging characteristics alone and that necessitate further
surgical exploration for final diagnosis. This definition applies to most
solid renal masses without fatty content and all cystic Bosniak type 3 and
type 4 lesions [1,
2].
The primary goal of this study was to evaluate the diagnostic performance
of state-of-the-art MDCT and MRI for differentiation of surgical from
nonsurgical kidney lesions. The techniques were compared with regard to image
quality and definite classification of the lesions.
Subjects and Methods
Patients
Between July 2002 and April 2003, 75 patients arrived in the department of
urology at our institution with kidney lesions that had been detected with
sonography and required further evaluation by imaging, operation, or both.
Twenty-eight of the patients (eight women, 20 men; age range, 27-84 years;
mean age, 64.8 ± 13.9 years [SD]) fulfilled the inclusion criteria and
were examined in this prospective study. The inclusion criteria were presence
of a kidney lesion that could not be classified with sonography alone and that
required further diagnostic evaluation, a minimum age of 18 years, and the
ability to give written informed consent. Exclusion criteria were pregnancy,
previous CT examination of sufficient quality (helical CT, slice thickness
less than 8 mm, unenhanced and delayed scans [60-120 sec postinjection]
covering the entire kidneys), and contraindications to iodinated contrast
agents or to MRI examination. Patients were also excluded from the study when
kidney lesions were present that could be confidently classified on the basis
of sonographic findings alone or when follow-up by sonography was considered
sufficient (simple cyst, homogeneous hyperechoic lesion). The interval between
MRI and MDCT examinations was less than 3 weeks. All patients gave written
informed consent for the study protocol, which had been approved by the
institutional ethics committee.
A total of 44 lesions had been detected sonographically. Nineteen (43.2%)
of the lesions were predominantly solid lesions, 12 (27.3%) were simple cystic
lesions larger than 15 mm, 11 (25.0%) were complex cystic lesions, and two
(4.5%) were inhomogeneous hyperechoic solid lesions. A total of 56 kidney
lesions were detected with MDCT and MRI. The final diagnoses were 16 renal
cell carcinomas, 15 Bosniak type 1 cysts larger than 15 mm, nine Bosniak type
2 cysts, nine oncocytomas, three angiomyolipomas, three inflammatory lesions,
and one adenoma. Simple Bosniak type 1 cysts smaller than 15 mm were not
included in the evaluation.
Eighteen patients underwent surgery on the basis of reported imaging
findings. Histologic examination revealed 13 renal cell carcinomas (nine clear
cell, three papillary, one chromophobe) and one angiomyolipoma in 13 patients;
nine oncocytomas in two patients (one patient with four oncocytomas in one
kidney and one patient with three oncocytomas in the left and two oncocytomas
in the right kidney); inflammatory lesions in two patients; and an adenoma in
one patient. Six Bosniak type 1 cysts larger than 15 mm were also found in the
resection specimen.
In three patients with very large inoperable renal cell carcinoma, the
final diagnosis was established at autopsy. Therefore 35 lesions were
confirmed at histologic examination. Confirmation of the remaining 21 lesions
was based on clinical follow-up evaluation (all cases) and imaging follow-up
examinations 6 and 12 months after the primary sonographic diagnosis (two
angiomyolipomas, six Bosniak type 1 cysts, and one case of inflammation) or
helical CT (nine Bosniak type 2 cysts and three type 1 cysts).
Imaging Technique: MDCT
All studies were performed with a 16-MDCT scanner (Somatom Sensation 16,
Siemens Medical Solutions). A triphasic imaging protocol consisting of an
unenhanced phase, an arterial phase, and a portal venous phase was used. All
scans were obtained with 16 x 0.75 mm collimation at 120 kV and 200 mAs
with the CareDose mode (Siemens Medical Solutions). The unenhanced and
arterial phases included the kidneys only, and the portal venous phase
included the entire abdomen. For better delineation of intestinal structures,
oral contrast medium was used in cases in which malignancy was strongly
suspected (n = 20; 30 mL of meglumine ioxithalamate [Telebrix,
Guerbet] in 1,000 mL of water 60 min before scanning). After the unenhanced
scan was obtained, 20 mL of contrast medium was administered IV to enhance the
renal pelvis and ureters, which were imaged with low-dose control scans at the
level of the ureteropelvic junction after 3-5 min. After opacification of the
collection system, 120 mL of iomeprol contrast medium (Imeron 300, Altana) was
administered at a flow rate of 4-5 mL/sec followed by a saline flush of 30 mL
at the same flow rate as the contrast medium (split-bolus technique). For the
arterial phase, the CareBolus technique was used. This technique allows
automatic start of the scan after a certain level of opacification (>120 H)
is reached in a region of interest placed by the user in the aorta at the
level of the renal arteries. After 80 sec, the portal venous scan of the
abdomen was started. All scans were obtained in deep inspiration. Each phase
was reconstructed with a slice thickness of 4 mm and a reconstruction
increment of 4 mm with a B30 kernel for filming on hard copies. In addition,
each phase was reconstructed with a 0.75-mm slice thickness and 0.7-mm
reconstruction increment (B30 kernel) for 3D applications. Semicoronal
re-constructions parallel to the kidneys in the portal venous phase with a
slice thickness of 4 mm were routinely obtained. All data were stored on
optical disks and transferred to a Leonardo workstation (Siemens Medical
Solutions) for further 3D reconstruction and image analysis.
View larger version (123K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3A —54-year-old man with complicated cystic lesion on right side not
correctly classified with MRI or CT in receiver operating characteristic (ROC)
analysis. MR images show small hyperintense cortical lesion (arrows)
not as intense as fluid compared with renal collecting system. No substantial
contrast enhancement is present in T1-weighted contrast-enhanced images with
fat suppression. T2-weighted sequence with fat saturation
|
|
View larger version (147K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3B —54-year-old man with complicated cystic lesion on right side not
correctly classified with MRI or CT in receiver operating characteristic (ROC)
analysis. MR images show small hyperintense cortical lesion (arrows)
not as intense as fluid compared with renal collecting system. No substantial
contrast enhancement is present in T1-weighted contrast-enhanced images with
fat suppression. T1-weighted axial image, late venous phase.
|
|
View larger version (162K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3C —54-year-old man with complicated cystic lesion on right side not
correctly classified with MRI or CT in receiver operating characteristic (ROC)
analysis. MR images show small hyperintense cortical lesion (arrows)
not as intense as fluid compared with renal collecting system. No substantial
contrast enhancement is present in T1-weighted contrast-enhanced images with
fat suppression. T1-weighted dynamic coronal image, venous phase.
|
|
View larger version (175K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3D —54-year-old man with complicated cystic lesion on right side not
correctly classified with MRI or CT in receiver operating characteristic (ROC)
analysis. CT scans corresponding to A-C show lesion (arrows)
is mean 20 H in unenhanced scan (D) but no measurable enhancement in
axial (E) or coronal (F) view of venous scan. Peripheral
enhancement could not be excluded, which led to classification of level 3 (not
sure) in ROC analysis and Bosniak classification of 2f. Histologic examination
of surgical specimen revealed small clear cell carcinoma.
|
|
View larger version (169K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3E —54-year-old man with complicated cystic lesion on right side not
correctly classified with MRI or CT in receiver operating characteristic (ROC)
analysis. CT scans corresponding to A-C show lesion (arrows)
is mean 20 H in unenhanced scan (D) but no measurable enhancement in
axial (E) or coronal (F) view of venous scan. Peripheral
enhancement could not be excluded, which led to classification of level 3 (not
sure) in ROC analysis and Bosniak classification of 2f. Histologic examination
of surgical specimen revealed small clear cell carcinoma.
|
|
View larger version (175K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3F —54-year-old man with complicated cystic lesion on right side not
correctly classified with MRI or CT in receiver operating characteristic (ROC)
analysis. CT scans corresponding to A-C show lesion (arrows)
is mean 20 H in unenhanced scan (D) but no measurable enhancement in
axial (E) or coronal (F) view of venous scan. Peripheral
enhancement could not be excluded, which led to classification of level 3 (not
sure) in ROC analysis and Bosniak classification of 2f. Histologic examination
of surgical specimen revealed small clear cell carcinoma.
|
|
Imaging Technique: MRI
MRI was performed with a 1.5-T superconducting imaging system (Gyroscan
Intera, Philips Medical Systems) equipped with a gradient system with an
amplitude of 23 mT/m in 200 µsec. A phased-array surface coil was centered
over the patient's abdomen at the level of the kidneys. The following
sequences were used: T2-weighted fast spin-echo sequence (TR/TE, 2,200/120;
echo-train length, 23; number of signals averaged, 4) in the axial plane with
spectral presaturation by inversion recovery and a slice thickness of 6 mm (no
spacing; respiratory triggering); T1-weighted unenhanced fast spin-echo
sequences (500/15; echo-train length, 3; number of signals averaged, 4) with
and without spectral presaturation by inversion recovery, slice thickness of 6
mm (no spacing; phase-encoding artifact reduction); dynamic
gadolinium-enhanced T1-weighted 2D gradient echo sequence (233/5.6; flip
angle, 80°, imaging time, 20 sec; breath-hold in expiration) in the
coronal plane, slice thickness of 5 mm at 0, 20, 50, and 160 sec after
administration of gadopentetate dimeglumine (Magnevist, Schering) 0.1 mmol/kg
body weight IV at a flow rate of 2 mL/sec. The T1-weighted fast spin-echo
sequences were repeated after injection of gadolinium.
For all images, parallel imaging was applied with sensitivity encoding at a
factor of 2. Sequences were repeated by the technical assistants if there were
motion artifacts. A radiologist on site checked the images for quality before
the patient left the imager. Image data were saved on optical disks and
transferred to a Leonardo workstation for further analysis.
View larger version (119K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 4A —55-year-old man with histologically confirmed multiple oncocytomas
in both kidneys. All were considered surgical lesions. Sonogram shows one
solid slightly hyperechoic mass (arrows) in left kidney.
|
|
View larger version (138K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 4B —55-year-old man with histologically confirmed multiple oncocytomas
in both kidneys. All were considered surgical lesions. MDCT scan shows solid
mass (arrow) on left side and two small cortical solid enhancing
lesions (arrows) on right side.
|
|
View larger version (134K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 4C —55-year-old man with histologically confirmed multiple oncocytomas
in both kidneys. All were considered surgical lesions. MR image of right
kidney shows small cortical solid enhancing lesion (arrow).
|
|
Image Analysis
All images were interpreted retrospectively by two experienced radiologists
in consensus without knowledge of clinical or histologic data. Both reviewers
had more than 3 years of experience in genitourinary imaging and were equally
trained in reading CT and MR images. CT and MR images were analyzed with a
Leonardo workstation in different sessions with at least 8 weeks between
sessions. If the two reviewers disagreed, a third fully trained faculty
radiologist with more than 3 years of experience in genitourinary imaging was
consulted, and the final decision was based on his vote.
The reviewers assessed overall image quality on a four-point scale from 0
to 3. Grade 0 indicated a diagnosis could not be made. Grade 1 indicated poor
image quality; a diagnosis could be made, but severe image artifacts were
present (the suspected lesion was seen and a differential diagnosis was
established, but some aspects, such as lymph nodes and renal vein thrombosis,
could not be properly analyzed). Grade 2 indicated good image quality with
only minor image artifacts (e .g., breathing artifacts); all important aspects
could be analyzed. Grade 3 indicated excellent image quality; no
artifacts.
For differentiation of surgical from nonsurgical lesions, receiver
operating characteristic (ROC) analysis was performed with five levels of
confidence. Level 1 indicated definitely nonsurgical lesion; level 2, probably
nonsurgical lesion; level 3, not sure; level 4, probably surgical lesion;
level 5, definitely surgical lesion. All Bosniak types 1 and 2 benign cysts,
angiomyolipomas, and areas of inflammation were considered nonsurgical lesions
because they can be considered benign solely because of their imaging
characteristics. All Bosniak types 3 and 4 cysts and all solid, enhancing
lesions (except angiomyolipoma) were considered surgical lesions (including
oncocytomas), because malignancy cannot be excluded on the basis of imaging
characteristics alone. The threshold for pathologic enhancement was set at a
10-H difference between unenhanced and venous scans.
Reviewers were required to assign a definite diagnosis to each lesion. If
histopathologic or follow-up examination confirmed the diagnosis in cases of
malignant lesions, the finding was rated true-positive. If histopathologic or
follow-up examination confirmed the diagnosis in cases of benign lesions, the
finding was rated true-negative. If histopathologic or follow-up examination
revealed a malignant instead of a suspected benign lesion, the finding was
rated false-negative. If histopathologic or follow-up examination revealed a
benign instead of a suspected malignant lesion, the finding was rated
false-positive. The 1997 International Union Against Cancer classification was
used for staging.
Statistical Analysis
ROC analysis was performed for MDCT and MRI separately to determine the
diagnostic performance of both techniques in differentiation of surgical from
nonsurgical lesions. Area under the curve (Az) was
calculated by logistic regression analysis. The optimal cutoff value was
defined as the cutoff value that maximized sensitivity and specificity and was
determined for each ROC curve separately. The corresponding 95% confidence
interval (CI) was obtained for each sensitivity and specificity value at each
level of confidence. Differences between ROC curves were tested on the basis
of 95% CI of the difference of the Az values. For
comparison of image quality, a paired Student's t test was used. All
statistical tests were performed at the 5% level of statistical significance
with the Stat-View program version 5.0 (SAS Institute, 1992-1998) or MedCalc
version 6.15.000 (Med-Calc, 1993-2002).
View larger version (154K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 6A —68-year-old woman with clear cell carcinoma of right kidney. MDCT
scans (A-D) have excellent image quality (grade 3) with clear depiction
of borders of tumor (arrow), especially in arterial phase (A,
C). MR images (E, F) have good image quality (grade 2), showing
tumor (arrow) with few motion artifacts. Axial arterial phase MDCT
scan.
|
|
View larger version (154K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 6B —68-year-old woman with clear cell carcinoma of right kidney. MDCT
scans (A-D) have excellent image quality (grade 3) with clear depiction
of borders of tumor (arrow), especially in arterial phase (A,
C). MR images (E, F) have good image quality (grade 2), showing
tumor (arrow) with few motion artifacts. Axial portal venous phase
MDCT scan.
|
|
View larger version (162K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 6C —68-year-old woman with clear cell carcinoma of right kidney. MDCT
scans (A-D) have excellent image quality (grade 3) with clear depiction
of borders of tumor (arrow), especially in arterial phase (A,
C). MR images (E, F) have good image quality (grade 2), showing
tumor (arrow) with few motion artifacts. Coronal arterial phase MDCT
scan.
|
|
View larger version (161K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 6D —68-year-old woman with clear cell carcinoma of right kidney. MDCT
scans (A-D) have excellent image quality (grade 3) with clear depiction
of borders of tumor (arrow), especially in arterial phase (A,
C). MR images (E, F) have good image quality (grade 2), showing
tumor (arrow) with few motion artifacts. Coronal portal venous phase
MDCT scan.
|
|
View larger version (130K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 6E —68-year-old woman with clear cell carcinoma of right kidney. MDCT
scans (A-D) have excellent image quality (grade 3) with clear depiction
of borders of tumor (arrow), especially in arterial phase (A,
C). MR images (E, F) have good image quality (grade 2), showing
tumor (arrow) with few motion artifacts. T2-weighted spectral
presaturation by inversion recovery axial MR image.
|
|
View larger version (166K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 6F —68-year-old woman with clear cell carcinoma of right kidney. MDCT
scans (A-D) have excellent image quality (grade 3) with clear depiction
of borders of tumor (arrow), especially in arterial phase (A,
C). MR images (E, F) have good image quality (grade 2), showing
tumor (arrow) with few motion artifacts. T1-weighted
gradient-refocused echo coronal MR image.
|
|
View larger version (117K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 7A —74-year-old woman with clear cell carcinoma of right kidney.
Excellent image quality (grade 3) in both MR images and MDCT scans. Large
hypervascular tumor and Bosniak type 1 cysts were correctly evaluated with
both techniques, and results were histologically confirmed. Sonogram shows
tumor (arrows) at upper pole.
|
|
View larger version (96K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 7B —74-year-old woman with clear cell carcinoma of right kidney.
Excellent image quality (grade 3) in both MR images and MDCT scans. Large
hypervascular tumor and Bosniak type 1 cysts were correctly evaluated with
both techniques, and results were histologically confirmed. Sonogram shows
cystic lesion (arrow) at lower pole.
|
|
View larger version (113K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 7C —74-year-old woman with clear cell carcinoma of right kidney.
Excellent image quality (grade 3) in both MR images and MDCT scans. Large
hypervascular tumor and Bosniak type 1 cysts were correctly evaluated with
both techniques, and results were histologically confirmed. MR images show
solid T2-weighted hyperintense (C) and contrast-enhancing lesion
(D, unenhanced T1-weighted MRI; E, contrast-enhanced T1-weighted
MRI) at upper pole and two cystic lesions without contrast enhancement at
lower pole of kidney.
|
|
View larger version (97K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 7D —74-year-old woman with clear cell carcinoma of right kidney.
Excellent image quality (grade 3) in both MR images and MDCT scans. Large
hypervascular tumor and Bosniak type 1 cysts were correctly evaluated with
both techniques, and results were histologically confirmed. MR images show
solid T2-weighted hyperintense (C) and contrast-enhancing lesion
(D, unenhanced T1-weighted MRI; E, contrast-enhanced T1-weighted
MRI) at upper pole and two cystic lesions without contrast enhancement at
lower pole of kidney.
|
|
View larger version (147K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 7E —74-year-old woman with clear cell carcinoma of right kidney.
Excellent image quality (grade 3) in both MR images and MDCT scans. Large
hypervascular tumor and Bosniak type 1 cysts were correctly evaluated with
both techniques, and results were histologically confirmed. MR images show
solid T2-weighted hyperintense (C) and contrast-enhancing lesion
(D, unenhanced T1-weighted MRI; E, contrast-enhanced T1-weighted
MRI) at upper pole and two cystic lesions without contrast enhancement at
lower pole of kidney.
|
|
View larger version (60K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 7F —74-year-old woman with clear cell carcinoma of right kidney.
Excellent image quality (grade 3) in both MR images and MDCT scans. Large
hypervascular tumor and Bosniak type 1 cysts were correctly evaluated with
both techniques, and results were histologically confirmed. MDCT scans show
solid lesion at upper pole and two cystic lesions at lower pole with equal
image quality compared to MRI (F, contrast-enhanced axial plane;
G, unenhanced and H, enhanced coronal reformations).
|
|
View larger version (116K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 7G —74-year-old woman with clear cell carcinoma of right kidney.
Excellent image quality (grade 3) in both MR images and MDCT scans. Large
hypervascular tumor and Bosniak type 1 cysts were correctly evaluated with
both techniques, and results were histologically confirmed. MDCT scans show
solid lesion at upper pole and two cystic lesions at lower pole with equal
image quality compared to MRI (F, contrast-enhanced axial plane;
G, unenhanced and H, enhanced coronal reformations).
|
|
View larger version (142K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 7H —74-year-old woman with clear cell carcinoma of right kidney.
Excellent image quality (grade 3) in both MR images and MDCT scans. Large
hypervascular tumor and Bosniak type 1 cysts were correctly evaluated with
both techniques, and results were histologically confirmed. MDCT scans show
solid lesion at upper pole and two cystic lesions at lower pole with equal
image quality compared to MRI (F, contrast-enhanced axial plane;
G, unenhanced and H, enhanced coronal reformations).
|
|
Results
Assessment of image quality showed that all images were of diagnostic
quality. Six MR images were considered of poor quality with many artifacts
(grade 1), whereas none of the MDCT scans was of poor quality. Eighteen MR
images were of good quality (grade 2), and four images were of excellent
quality (grade 3). Six MDCT scans were of good image quality and 22 scans of
excellent image quality. The mean image quality grade for all MDCT scans was
2.79 ± 0.42. For the MR images, the mean image quality grade was 1.93
± 0.60. The difference between the image quality of MDCT and that of
MRI was statistically significant (p < 0.001). The results are
summarized in Figure 1.
In differentiation of surgical from nonsurgical kidney lesions, both
methods performed comparably. The ROC analysis resulted in
Az values of 0.979 for MDCT (95% CI, 0.948-1.009) and
0.957 for MRI (95% CI, 0.895-1.018) (Fig.
2). Both results were within the corresponding 95% CI, and the
differences between the ROC curves were not statistically significant
(p = 0.739). The cut-off value that optimized sensitivity and
specificity for MDCT was level 4, meaning that all lesions characterized as
level 4 (probably surgical lesions) and level 5 (definitely surgical lesion)
were considered surgical lesions. At this level, sensitivity was 92.3% and
specificity was 96.3%. For MRI, the optimal cutoff value was the same
confidence level as for MDCT, resulting in a sensitivity of 92.3% and a
specificity of 91.3%.
View larger version (154K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 8B —78-year-old man with clear cell carcinoma of right kidney and
invasion of renal vein and inferior vena cava. Axial portal venous phase MDCT
scan of excellent image quality (grade 3) clearly shows tumor
(arrow).
|
|
View larger version (141K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 8C —78-year-old man with clear cell carcinoma of right kidney and
invasion of renal vein and inferior vena cava. Axial arterial phase MDCT scan
in curved planar reformation shows tumor thrombus (arrow). Tumor
thrombus was missed on MR images because of severe motion artifacts. MDCT scan
has excellent image quality (grade 3).
|
|
In exact diagnosis of the lesions, MDCT correctly depicted 15 of 16 renal
cell carcinomas (sensitivity, 93.3%; Fig.
3A,
3B,
3C,
3D,
3E and
3F) and correctly depicted 26
of 38 benign lesions (specificity, 68.4%). The positive predictive value (PPV)
was 55.6% (15/27), and the negative predictive value (NPV) was 96.3% (26/27).
All oncocytomas were falsely classified as renal cell carcinomas (Fig.
4A,
4B,
4C and
4D). One inflammatory lesion
was falsely classified as carcinoma. All angiomyolipomas, two inflammatory
lesions, and one adenoma were correctly recognized as benign lesions (Fig.
5A,
5B,
5C and
5D). Two Bosniak type 2 cysts
smaller than 15 mm were falsely classified as Bosniak type 1 cysts on
MDCT.
MRI also correctly depicted 15 of 16 renal cell carcinomas (sensitivity,
93.3%) and correctly depicted 25 of 35 benign lesions (specificity, 71.4%;
Figs. 6A,
6B,
6C,
6D,
6E,
6F and
7A,
7B,
7C,
7D,
7E,
7F,
7G and
7H). The PPV was 55.6% (15/27)
and the NPV was 96.3% (26/27). Again, all oncocytomas were falsely classified
as carcinomas. One adenoma was misclassified as a malignant lesion. All
angiomyolipomas and inflammatory lesions were correctly classified. Three
benign lesions smaller than 10 mm (one 6-mm angiomyolipoma and oncocytomas
measuring 6 mm and 7 mm) were detected with MDCT but not with MRI. Two Bosniak
type 2 cysts smaller than 15 mm with small calcifications were falsely
classified as Bosniak type 1 cysts, because the calcifications were not
detected with MRI performed with our MRI protocol.
On a per-patient basis, the malignant lesions of 15 of 16 patients were
correctly classified with both MDCT and MRI, resulting in a sensitivity of
93.8% for both techniques. The benign lesions of eight of 12 patients were
correctly classified with both MDCT and MRI, resulting in a specificity of
66.7% for both techniques. The PPV and NPV for both MDCT and MRI on a
per-patient basis were 78.9% (15/19) and 88.9% (8/9).
TNM staging of surgically removed malignant lesions (n = 13) was
correct for both MDCT and MRI in eight of 13 lesions. MRI interpretation led
to overstaging of three lesions and understaging of two lesions. MDCT
interpretation led to overstaging of four lesions and understaging of one
lesion. Minimal invasion of renal veins in the hilus (stage T3a) was missed
with both techniques, and one of three surgically proven tumor thrombi in the
renal veins (stage T3b) was missed with MRI (Figs.
8A,
8B and
8C). Overstaging with both
techniques was always caused by false classification of enlarged lymph nodes
(> 15 mm) as malignant lymph nodes (N1). These nodes proved to be reactive
lymph nodes at histologic examination.
Discussion
In this study we compared the performance of 16-MDCT and MRI in the
characterization of kidney lesions. MDCT proved superior to MRI with regard to
image quality. Both MDCT and MRI proved excellent for differentiating surgical
from nonsurgical kidney lesions. However, in definite classification of
lesions, both MDCT and MRI had good sensitivity and NPV but limited
specificity and low PPV.
We focused on differentiation of surgical from nonsurgical renal lesions
because surgeons determining whether a patient needs an operation or
conservative care base the decision mainly on imaging findings. ROC analysis
showed no statistical difference in the performance of MDCT and MRI in
differentiation of surgical from nonsurgical renal lesions and revealed
sensitivity and specificity of 92.3% and 96.3% for MDCT and 92.3% and 91.3%
for MRI. Concerning reviewer confidence, both methods performed equally and
had an optimal cutoff at a confidence level of 4. CT has been a commonly used
tool in the evaluation of kidney lesions for many years, reflected, for
example, in the Bosniak classification of cystic kidney lesions introduced in
1986 [1,
10]. This classification is
based on CT criteria [1,
2] but can also be applied to
MR images. In our study, this classification was used for evaluation of cystic
renal masses, which are in many cases the most difficult lesions to
evaluate.
Similar to the approach in our study, that used by Israel et al.
[11] was a comparison of CT
(helical CT with single-detector to four-slice scanners) and MRI in a study
involving 59 patients with cystic renal masses classified by the Bosniak
system. CT and MRI findings were similar in most of the cystic masses, but in
10% of cases, the MRI findings led to upgrading of Bosniak type. Our study
corroborated this finding in that we found both MDCT and MRI had excellent
sensitivity and specificity in differentiation of surgical from nonsurgical
kidney lesions. However, we had no cases in which the MRI findings led to
upgrading of lesions. This finding may be due to differences in the patient
populations, because we evaluated not only cystic but also solid renal
lesions. We consequently had fewer complicated cystic lesions in our
evaluation, which may explain in part the differences in study outcome. In our
study, two large (> 3 cm) cystic carcinomas without contrast enhancement
exhibited heterogeneity of cyst contents on MDCT and on MRI, so both
techniques provided the same information. Another factor may be that all
patients in our study underwent 16-MDCT, which had excellent image quality
with high resolution in all image planes. Although there is no direct proof of
this hypothesis, Israel et al. also mention as a limitation to their study
that only axial CT scans were analyzed and that the multiplanar MR images
might have been responsible for the better performance of MRI. A study by
Tello et al. [12] emphasized
the importance of heterogeneous signal intensity in malignant renal lesions
without contrast enhancement on MRI compared with CT. Therefore it cannot be
excluded that in some cases of cystic lesions without contrast enhancement,
inhomogeneity of the lesion on MRI may lead to an upgrade from nonsurgical
(Bosniak 2) to surgical lesion (Bosniak 3).
To the best of our knowledge only two studies have compared
state-of-the-art helical CT and MRI in evaluation of renal lesions
[6,
8]. In the study by Walter et
al. [6], all histologically
proven masses were successfully detected with both MRI and CT. The two
techniques also had equal performance in the characterization and staging of
renal lesions. These results corresponded to our findings. However, the study
by Walter et al. focused mainly on solid renal lesions, and the authors noted
that there were no complicated renal cysts in the patient sample. Complicated
renal cysts are the most interesting lesions for radiologists, because in many
cases the imaging findings have a direct influence on patient management. In
their study, Hallscheidt et al.
[8] also compared MDCT and MRI
but only for staging of renal cell carcinoma and not for characterization of
kidney lesions.
Our examinations not only corroborated the findings of the other studies
but also contribute to the growing literature on evaluation of various renal
lesions, that is, solid and cystic lesions, including complicated cysts.
Moreover, to our knowledge, this study is the first comparison of 16-MDCT and
1.5-T MRI in the evaluation of renal masses in the same patient group. This
factor is important because dramatic improvement in the quality of CT since
the introduction of MDCT has led to high accuracy in the preoperative
evaluation of renal cell carcinoma
[13]. The image quality of
MDCT in our study was significantly better than that of MRI. Classification
and delineation of small cortical tumors and large tumors were equally good or
better with MDCT compared with MRI. MDCT also delineated all tumor thrombi in
the renal veins and vena cava, whereas one thrombus was missed on MRI because
of the presence of motion artifacts. In addition to the high spatial
resolution of MDCT and the capability for high-quality reconstructions, an
explanation for this result may be the shorter examination time of MDCT
compared with MRI. This explanation remains speculative because we did not
perform a systematic analysis of the influence of acquisition time on image
quality. The anatomic detail achieved with MDCT is, for example, of paramount
importance for planning nephron-sparing renal surgery
[14,
15], although this feature
applies not only to 16-MDCT but also to 4-MDCT. The relation of the tumor to
adjacent vessels and to the renal collecting system was well visualized with
MDCT because of the split-bolus technique we used for opacification of the
renal collecting system and because of the use of multiplanar reformations.
The split-bolus technique also obviates a separate urographic scan.
Consequently, some advantages of MRI over incremental or single-detector CT,
such as high-quality multiplanar imaging and better delineation of tumor
thrombus [16], may no longer
apply with the use of MDCT.
We evaluated the performance of MDCT and MRI in differentiation of benign
from malignant lesions. Although the sensitivity was high at 93.8% for both
MDCT and MRI, specificity was quite low at 68.4% and 71.4%, respectively,
values reflected in low PPVs. Both techniques were very good in detection of
malignant renal lesions, and the probability of misclassifying a malignant
lesion as benign is low. Many lesions managed surgically, however, turned out
to be benign after the operation. Although this result may have been due in
part to the large number of oncocytomas in our study (two patients with nine
oncocytomas), it reflects a common problem in renal imaging: Solid enhancing
lesions (without fat) must be considered surgical lesions, because no definite
imaging criteria exist for differentiating carcinomas from adenomas and
oncocytomas
[17-19].
There were limitations to our study. The image analysis was done in
consensus, and therefore interobserver variability was not evaluated. In
addition, the sample size was relatively small, and pathologic correlation was
not available in all cases. Although follow-up findings were analyzed for the
lesions for which there was no pathologic correlation, the follow-up time was
not longer than 12 months, because 16-MDCT had been only recently introduced
in our department. Therefore some Bosniak type 2 cysts may turn out to be
malignant, although this outcome is not likely, according to the literature
[20-22].
We also did not include renal masses seen incidentally at CT or MRI
examinations. Exclusion of these lesions may have affected our specificity,
because more small and fewer clearly solid cancers probably would have been
included. Moreover, we included both sonographically solid and sonographically
cystic renal lesions in our study. Therefore it cannot be excluded that the
results may be different in a population with predominantly cystic lesions,
because MRI may be superior to CT in the detection of inhomogeneities in
cystic lesions [11].
In conclusion, both 16-MDCT and MRI perform excellently in differentiation
of surgical from nonsurgical renal lesions that have been detected with
sonography. However, neither technique can be used to further differentiate
benign from malignant tumors in cases of solid enhancing lesions without fat.
Because it has a shorter acquisition time, wider availability, and is less
expensive than MRI, MDCT continues to be the preferred technique for further
evaluation of sonographically detected renal masses, especially when
malignancy is already strongly suspected, as in solid heterogeneous masses.
However, MRI is a useful alternative in cases of contraindications to use of
iodinated contrast agents or when radiation dose is an issue, as in imaging of
young patients, repeated follow-up examinations, and imaging during
pregnancy.
References
- Masood J, Lane T, Koye B, Vandal MT, Barua JM, Hill JT. Renal cell
carcinoma: incidental detection during routine ultrasonography in men
presenting with lower urinary tract symptoms. BJU Int2001; 88:671
-674[Medline]
- Silverman SG, Lee BY, Seltzer SE, et al. Small (< 3 cm) renal
masses: correlation of spiral CT features and pathologic findings.
AJR 1994; 163:597
-605[Abstract/Free Full Text]
- Zagoria RJ, Bechtold RE, Dyer RB. Staging of renal adenocarcinoma:
role of various imaging procedures. AJR1995; 164:363
-370[Abstract/Free Full Text]
- Semelka RC, Hricak HH, Stevens S, Finegold R, Tomei E, Carrol PR.
Combined gadolinium-enhanced and fat-saturation MR imaging of renal masses.
Radiology 1991;178
: 803-809[Abstract/Free Full Text]
- Kopka L, Fischer U, Zoeller G, et al. Dual phase helical CT of the
kidney: value of the corticomedullary and nephrographic phase for evaluation
of renal lesions and preoperative staging of renal cell carcinoma.
AJR 1997; 169:1573
-1578[Abstract/Free Full Text]
- Walter C, Kruessell M, Gindele A, Brochhagen HG, Gossmann A,
Landwehr P. Imaging of renal lesions: evaluation of fast MRI and helical CT.
Br J Radiol 2003;76
: 696-703[Abstract/Free Full Text]
- Hallscheidt P, Stolte E, Roeren T, Pomer S, Drehmer I, Kauffmann
GW. The staging of renal-cell carcinomas in MRT and CT: a prospective
histologically controlled study. Rofo1998; 168:165
-170[Medline]
- Hallscheidt PJ, Bock M, Riedasch G, et al. Diagnostic accuracy of
staging renal cell carcinomas using multidetector-row computed tomography and
magnetic resonance imaging: a prospective study with histopathologic
correlation. J Comput Assist Tomogr 2004;28
: 333-339[CrossRef][Medline]
- Bosniak MA. The current radiological approach to renal cysts.
Radiology 1986;158
: 1-10[Abstract/Free Full Text]
- Dunnick NR, Sandler CM, Newhouse JH, Amis ES Jr.
Textbook of uroradiology, 3rd ed. Philadelphia, PA:
Lippincott 2001: 123-149
- Israel GM, Hindman N, Bosniak MA. Evaluation of cystic renal
masses: comparison of CT and MR imaging by using the Bosniak classification
system. Radiology 2004;231
: 365-371[Abstract/Free Full Text]
- Tello R, Davison BD, ÓMalley M, et al. MR imaging of renal
masses interpreted on CT to be suspicious. AJR2000; 174:1017
-1022[Abstract/Free Full Text]
- Catalano C, Frailoi F, Laghi A, et al. High-resolution
multidetector CT in the preoperative evaluation of patients with renal cell
carcinoma. AJR 2003;180
: 1271-1277[Abstract/Free Full Text]
- Smith PA, Marshall FF, Corl FM, Fishman EK. Planning
nephron-sparing renal surgery using 3 D helical CT angiography. J
Comput Assist Tomogr 1999;23
: 649-654[CrossRef][Medline]
- Hallscheidt P, Schoenberg S, Schenk JP, Zuna I, Petirsch O,
Riedasch G. Multi-slice CT in the planning of nephron-sparing interventions
for renal cell carcinoma: prospective study correlated with histopathology.
Rofo 2002; 174:898
-903[Medline]
- Pretorius ES, Wickstrom ML, Siegelman ES. MR imaging of renal
neoplasms. Magn Reson Imaging Clin N Am2000; 8:813
-836[Medline]
- Kodama K, Nagano K, Akimoto M, Suzuki S. Small renal oncocytoma
with central cystic degeneration. Int J Urol2004; 11:110
-113[Medline]
- Li G, Guilleron M, Gentil-Perret A, Tostain J. Characteristics of
image-detected solid renal masses: implication for optimal treatment.
Int J Urol 2004;11
: 63-67[CrossRef][Medline]
- Schatz SM, Lieber MM. Update on oncocytoma. Curr Urol
Rep 2003; 4:30
-35[Medline]
- Bosniak MA. Diagnosis and management of patients with complicated
cystic lesions of the kidney. AJR 1997;169
: 819-821[Free Full Text]
- Bosniak MA. The use of the Bosniak classification system for renal
cysts and cystic tumors. J Urol 1997;157
: 1852-1853[CrossRef][Medline]
- Kreft B, Schild HH. Cystic renal lesions.
Rofo 2003; 175:892
-903[Medline]
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
|
|
 
C. S. Ng, C. G. Wood, P. M. Silverman, N. M. Tannir, P. Tamboli, and C. M. Sandler
Renal Cell Carcinoma: Diagnosis, Staging, and Surveillance
Am. J. Roentgenol.,
October 1, 2008;
191(4):
1220 - 1232.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
M. F. Meloni, M. Bertolotto, C. Alberzoni, S. Lazzaroni, C. Filice, T. Livraghi, and G. Ferraioli
Follow-Up After Percutaneous Radiofrequency Ablation of Renal Cell Carcinoma: Contrast-Enhanced Sonography Versus Contrast-Enhanced CT or MRI
Am. J. Roentgenol.,
October 1, 2008;
191(4):
1233 - 1238.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
