Hematuria Portal Venous Phase MultiDetector Row CT of the

Document Sample
Hematuria Portal Venous Phase MultiDetector Row CT of the Powered By Docstoc
					                                          Note: This copy is for your personal non-commercial use only. To order presentation-ready
                                          copies for distribution to your colleagues or clients, contact us at www.rsna.org/rsnarights.

                                                                                   Hematuria: Portal Venous Phase
                                                                                   Multi–Detector Row CT of the
                                                                                   Bladder—A Prospective Study1

                         Sung Bin Park, MD
                                                                                        Purpose:    To prospectively determine the accuracy of portal venous
                         Jeong Kon Kim, MD
                                                                                                    phase helical multi– detector row computed tomography
                         Hyun Joo Lee, MD
                                                                                                    (CT) for bladder lesion evaluation in patients with hematu-
                         Hyuck Jae Choi, MD                                                         ria by using cystoscopy as the reference standard.
                         Kyoung-Sik Cho, MD
                                                                                    Materials and   The study was approved by the institutional review board
                                                                                        Methods:    for human investigation, and informed consent was ob-
                                                                                                    tained from all patients. This study included 118 patients
                                                                                                    (91 male, 27 female; age range, 15– 87 years; mean age
                                                                                                    standard deviation, 62 years 14) who underwent portal
                                                                                                    venous phase multi– detector row CT (scanning delay, 70
                                                                                                    seconds; section thickness, 2 mm) and conventional cys-
                                                                                                    toscopy because of painless gross hematuria or recurrent
                                                                                                    microscopic hematuria. Two reviewers with different ex-
                                                                                                    perience levels independently evaluated the bladder for
                                                                                                    lesions at CT in a prospective fashion. The statistic was
                                                                                                    used to determine the per lesion and per patient agree-
                                                                                                    ment between the two reviewers and between the CT
                                                                                                    and cystoscopic findings. The sensitivity and specificity of
                                                                                                    multi– detector row CT for bladder lesion detection were
                                                                                                    analyzed for numbers of lesions and for numbers of pa-

                                                                                         Results:   Multi– detector row CT showed excellent per lesion (
                                                                                                    0.839) and per patient (     0.881) agreement between the
                                                                                                    two reviewers. Respective per lesion and per patient
                                                                                                    agreement between the CT and cystoscopic findings was
                                                                                                    also excellent in the first (     0.866 and      0.881) and
                                                                                                    second (     0.802 and       0.863) reviewers. The sensitiv-
                                                                                                    ity and specificity of multi– detector row CT were 89%–
                                                                                                    92% and 88%–97%, respectively, in the per lesion analysis
                                                                                                    and 95% and 91%–93%, respectively, in the per patient
                                                                                                    analysis for both reviewers. All statistical parameters of
                                                                                                    diagnostic accuracy were similar between the two review-
                                                                                                    ers (P .05).

                                                                                      Conclusion:   Portal venous phase multi– detector row CT can provide
                                                                                                    high accuracy and reader agreement for bladder lesion
                                                                                                    detection in patients with painless gross hematuria and
                                                                                                    recurrent microscopic hematuria; these results indicate
                           From the Department of Radiology, Asan Medical Cen-
                                                                                                    that multi– detector row CT can be used as the initial
                         ter, University of Ulsan, 388-1 Poongnap-dong, Songpa-
                         gu, Seoul 138-736, South Korea (S.B.P., J.K.K., H.J.L.,                    bladder examination in such patients.
                         K.S.C.); and Department of Radiology, National Cancer
                         Center, Goyang, South Korea (H.J.C.). Received June 18,                     RSNA, 2007
                         2006; revision requested August 21; revision received
                         November 7; accepted December 18; final version ac-
                         cepted April 9, 2007. Address correspondence to
                         J.K.K. (e-mail: rialto@amc.seoul.kr ).

                             RSNA, 2007

                         798                                                                                                   Radiology: Volume 245: Number 3—December 2007
GENITOURINARY IMAGING: Hematuria: Portal Venous Phase CT of the Bladder                                                                        Park et al

        ladder lesions account for 15% of       seem to be limited in proving the use-      maining 28 patients were enrolled in
        all causes of hematuria (1,2). Ac-      fulness of multi– detector row CT as a      our study.
        curate bladder evaluation is thus       primary bladder examination because             In the other 440 patients who un-
important when painless gross hematu-           they retrospectively evaluated patients     derwent the CT examination first, con-
ria or repeated microscopic hematuria           with bladder cancer already diagnosed       ventional cystoscopy was performed ex-
is noted in patients with negative upper        (3,9,10). Moreover, because the detec-      cept in patients who: (a) had urolithiasis
urinary tract findings.                          tion of bladder lesions may vary accord-    and normal bladder at multi– detector
     In patients at high risk for bladder       ing to the observer’s experience level,     row CT; (b) had renal parenchymal tu-
lesions, conventional cystoscopy has been       the reliability of multi– detector row CT   mors other than urothelial tumors but
used as the primary examination. How-           also should be evaluated. Thus, the pur-    normal bladder at multi– detector row
ever, this examination has drawbacks, in-       pose of our study was to prospectively      CT; (c) had normal bladder at multi–
cluding patient discomfort, equipment           determine the accuracy of portal venous     detector row CT and clinical or labora-
requirements, and relatively high cost.         phase multi– detector row helical CT for    tory signs suggestive of renal parenchy-
In addition, conventional cystoscopic           bladder lesion evaluation in patients       mal disease; or (d) had a contraindica-
examination can be used to evaluate             with hematuria by using cystoscopy as       tion for conventional cystoscopy, such
only mucosal lesions but not the mus-           the reference standard.                     as bacteriuria, acute cystitis, urethritis,
cular layer or surrounding structures                                                       prostatitis, or obstructive prostatic hy-
(2–6).                                                                                      pertrophy with normal upper urinary
     Multi– detector row computed to-             Materials and Methods                     tract at multi– detector row CT. Ac-
mography (CT) is an accurate examina-           This study was approved by the insti-       cording to these criteria, conventional
tion for bladder lesion detection. Re-          tutional review board of the Asan           cystoscopy was performed after CT in
sults of several studies show that CT has       Medical Center (Seoul, South Korea)         90 patients who were enrolled in our
satisfactory accuracy for bladder lesion        for human investigation. Informed con-      study. Consequently, 118 patients (91
detection (3,7–13); authors of a recent         sent was obtained from all patients after   male, 27 female; age range, 15– 87
study (7) reported the sensitivity and          the study details, including radiation      years; mean age      standard deviation,
positive predictive value of multi– detec-      dose information, were explained to         62 years 14) were included in the final
tor row CT to be 97% and 95%, respec-           them.                                       study group (Fig 1).
tively.                                                                                         The interval between multi– detec-
     Multi– detector row CT has advan-          Patients                                    tor row CT and conventional cystoscopy
tages over conventional cystoscopy, such        Between December 2004 and June 2005,        was 8 days 7 (range, 0 –29 days). This
as the higher level of patient tolerance for    471 patients underwent CT owing to          interval was 10 days or fewer in 73 pa-
CT and the ability to allow evaluation of       single or multiple episodes of painless     tients, 11–20 days in 28 patients, and
the entire urinary tract and surrounding        gross hematuria or recurrent micro-         21–29 days in 17 patients.
structures (14–17). Therefore, if multi–        scopic hematuria that had occurred
detector row CT could provide accuracy          more than three times in 1 month. All       Conventional Cystoscopy
and reliability comparable to those of          CT examinations were performed ac-          Conventional cystoscopy was performed
conventional cystoscopy, it would be            cording to a dedicated scanning proto-      by four staff urologists in our institution
beneficial for use as a primary bladder          col.                                        (Asan Medical Center) who had 10 –12
examination to provide a guideline for               Of the 471 patients, 31 who under-     years of experience in cystoscopy. In
establishing the ensuing diagnostic and         went conventional cystoscopy before         patients who underwent CT before con-
therapeutic strategies. Previous studies        CT were enrolled in our study. Three of
                                                these 31 patients were excluded be-
                                                cause they had undergone transurethral      Published online before print
 Advance in Knowledge                           resection for bladder tumor before the      10.1148/radiol.2452061060
    Portal venous phase multi– detec-           CT examination. Consequently, the re-       Radiology 2007; 245:798 – 805
    tor row CT offers high accuracy
    (sensitivity, 89%–92% in per le-                                                        Author contributions:
                                                                                            Guarantor of integrity of entire study, J.K.K.; study con-
    sion analysis and 95% in per pa-
    tient analysis; specificity, 88%–              Implication for Patient Care              cepts/study design or data acquisition or data analysis/
                                                                                            interpretation, all authors; manuscript drafting or manu-
    97% in per lesion analysis and                   In view of its accuracy, we believe    script revision for important intellectual content, all au-
    91%–93% in per patient analysis)                 portal venous phase multi– detec-      thors; approval of final version of submitted manuscript,
    and reader agreement (                           tor row CT can be used as an ini-      all authors; literature research, S.B.P., J.K.K., H.J.L.,
    0.802– 0.881 in two reviewers at                 tial bladder examination in pa-        K.S.C.; clinical studies, S.B.P., J.K.K., K.S.C.; experimental
    both per lesion and per patient                  tients with painless gross hematu-     studies, J.K.K.; statistical analysis, J.K.K.; and manuscript
                                                                                            editing, all authors
    analysis) for bladder lesion                     ria and recurrent microscopic
    detection.                                       hematuria.                             Authors stated no financial relationship to disclose.

Radiology: Volume 245: Number 3—December 2007                                                                                                        799
GENITOURINARY IMAGING: Hematuria: Portal Venous Phase CT of the Bladder                                                            Park et al

ventional cystoscopy, urologists were           mation of 16      0.75 mm, a table feed    sion at the start of our study. We differ-
aware of CT results. Bladder lesions            per rotation of 12.0 mm per rotation, an   entiated observer performance accord-
were considered present when conven-            effective section thickness of 2 mm, and   ing to experience by restricting the time
tional cystoscopy revealed papillary or         a reconstruction interval of 2 mm.         spent on CT image interpretation to 3
sessile lesions from the bladder wall,                                                     minutes per patient.
color change on the bladder mucosa, or          Image Analysis                                 Bladder lesions were considered
ulcerative lesions. In each patient with        All CT images were interpreted for         present when the CT images revealed
positive cystoscopic findings, we re-            patient care. However, to eliminate        hyperenhanced lesions that protruded
corded the location, maximum diame-             any possible bias that awareness of        into the bladder lumen or hyperen-
ter, and number of bladder lesions.             the upper urinary tract findings might      hanced wall thickening in comparison
                                                affect the bladder evaluation, the         with the rest of the bladder wall. In
CT Examination                                  presence of a bladder lesion was de-       each patient, the presence or absence
All CT data were obtained by using a            termined by using only bladder CT im-      of a bladder lesion was recorded inde-
16-section CT scanner (SOMATOM Sen-             ages for the purposes of our study.        pendently by the two reviewers. In pa-
sation 16; Siemens Medical Systems, Er-         Bladder CT images were saved sepa-         tients with positive CT findings, the
langen, Germany). Patients were in-             rately and were displayed on a picture     location and number of bladder le-
structed not to void for at least 1 l⁄2 hours   archiving and communication system         sions were also independently re-
before the examination to obtain ade-           (Petavision; Asan Medical Center,          corded. For the location of the bladder
quate bladder distension. All patients          Seoul, South Korea). Two genitouri-        lesion, we performed lesion-by-lesion
received 600 –900 mL of 2% barium sul-          nary radiologists, including an experi-    correlation between portal venous
fate suspension (E-Z CAT; E-Z-Em,               enced reviewer (J.K.K.) and a less-ex-     phase CT and cystoscopy.
Westbury, NY) orally 1 hour before CT           perienced reviewer (S.B.P.), both of
scanning.                                       whom were unaware of any patient infor-    Statistical Analysis
    The CT protocol of our institution          mation, the cystoscopic findings, and the   The weighted        statistic was used to
for evaluating patients with hematuria          CT findings of the upper urinary tract,     evaluate interobserver agreement be-
includes unenhanced scanning, cortico-          independently evaluated the bladder        tween the two reviewers with regard to
medullary phase scanning, portal ve-            CT images. The first reviewer (experi-      identifying bladder lesions by using both
nous phase scanning for the bladder,            enced reviewer) was a staff radiologist    per lesion and per patient analysis.
and early excretory phase scanning.             who had worked for 7 years in the geni-        The results of the CT image inter-
During unenhanced scanning, adequate            tourinary division, whereas the second     pretation and the conventional cysto-
bladder distension was monitored by a           reviewer (less-experienced reviewer)       scopic findings were also compared for
radiologist (H.J.L.) with 3 years of ex-        was a radiology fellow who had worked      the two reviewers by using the weighted
perience in pelvic CT; adequate blad-           for 4 months in the genitourinary divi-       statistic; their agreement was also
der distension was indicated when the
bladder was spherical and had a mini-
                                                  Figure 1
mum diameter greater than 8 cm. After
the unenhanced CT images were ob-                                                                        Figure 1: Flow diagram shows
tained, iopromide (Ultravist 300; Scher-                                                                 study protocol between December
ing, Berlin, Germany) or iopamidol (Io-                                                                  2004 and June 2005. MDCT
pamiro 300; Bracco, Milan, Italy) was                                                                    Multi– detector row helical CT.
administered intravenously into an an-
tecubital vein by using a power injector
at a dose of 2 mL per kilogram of body
weight at a rate of 3 mL/sec to a maxi-
mum of 160 mL. Portal venous phase
CT for the bladder covered the entire
bladder at the 70-second scanning de-
lay, which was determined on the basis
of the data from our previous study,
which indicated the optimal scanning
delay for bladder tumor detection (7).
    The CT parameters for bladder
scanning were a beam pitch of 3, an
x-ray tube voltage of 120 kV, a tube
current of 130 –150 mAs, a gantry rota-
tion speed of 0.5 second, a beam colli-

800                                                                                             Radiology: Volume 245: Number 3—December 2007
GENITOURINARY IMAGING: Hematuria: Portal Venous Phase CT of the Bladder                                                                     Park et al

evaluated by using both per lesion and                       test after adjusting for the clustering ef-   patients. In 11 patients with diffuse
per patient methods. A weighted                              fect (18).                                    bladder lesions, the number of lesions
value of less than 0.20 was considered                           All statistical analyses were per-        in each bladder was counted as one.
poor; 0.21– 0.40, fair; 0.41– 0.60, mod-                     formed by using statistical software          Thirty-five (34%) of 104 lesions were
erate; 0.61– 0.80, good; and 0.81–1.00,                      (SPSS, version 12.0.0; SPSS, Chicago,         less than 10 mm in maximum diameter.
excellent.                                                   IL), and significance was indicated when           The final diagnosis of these bladder
    By using the conventional cysto-                         the P value was less than .05.                lesions was transitional cell carcinoma
scopic findings as a standard of refer-                                                                     in 96 lesions, cystitis in seven lesions,
ence, we analyzed the sensitivity, speci-                                                                  and metastasis from gastric cancer in
ficity, positive predictive value, negative                    Results                                      one lesion; three of 96 transitional cell
predictive value, and overall accuracy of                                                                  carcinomas were carcinoma in situ.
portal venous phase multi– detector row                      Conventional Cystoscopy                       All cases of transitional cell carcinoma
CT for bladder lesion detection for each                     Conventional cystoscopy revealed 104          and metastasis were histologically con-
reviewer; all statistical parameters were                    bladder lesions in 63 (53%) of 118 study      firmed by means of transurethral bi-
evaluated by using both per lesion and                       patients. The number of lesions in each       opsy. Six of seven cases of cystitis were
per patient methods. Then each statisti-                     bladder was one in 43 patients, two in        histologically confirmed by means of
cal parameter was compared for the                           11 patients, three in three patients,         transurethral biopsy, and the remaining
two reviewers by using the McNemar                           four in three patients, and six in three      case was confirmed by means of visual
                                                                                                           inspection at conventional cystoscopy.

   Table 1                                                                                                 Agreement between CT and Cystoscopic
                                                                                                           Findings and between Reviewers
   Agreement on the Number of Bladder Lesions in Each Patient
                                                                                                           Per lesion analysis.—Agreement between
    A: Agreement between Cystoscopic Findings and First Reviewer                                           CT image interpretation and cystoscopic
    No. of Lesions                                                 First Reviewer                          findings was noted in 108 (92%) of the
    at Cystoscopy            0             1             2               3          4   6       Total      118 patients by the first reviewer and in
                                                                                                           103 (87%) of the 118 patients by the
     0                       53             2            0              0           0   0        55        second reviewer (Table 1). The agree-
     1                        5            38            0              0           0   0        43        ment between CT image interpretation
     2                        1             1            9              0           0   0        11
                                                                                                           and cystoscopic findings was excellent
     3                        0             0            0              3           0   0         3
                                                                                                           for both the first (      0.866   0.041)
     4                        0             0            0              1           2   0         3
                                                                                                           and second (      0.802 0.047) review-
     6                        0             0            0              0           0   3         3
                                                                                                           ers. The two reviewers agreed on the
     Total                   59            41            9              4           2   3       118
                                                                                                           number of lesions in 106 (90%) of
    B: Agreement between Cystoscopic Findings and Second Reviewer
                                                                                                           the 118 patients, so the interobserver
    No. of Lesions                                             Second Reviewer
                                                                                                           agreement was excellent (      0.839
    at Cystoscopy            0             1            2            3              4   6       Total
                                                                                                           0.043) (Fig 2).
     0                       50             4            1               0          0   0        55            Per patient analysis.—Because agree-
     1                        3            38            2               0          0   0        43        ment (Table 2) between CT findings and
     2                        0             3            8               0          0   0        11        cystoscopic findings was noted in 111
     3                        0             0            1               2          0   0         3        (94%) of the 118 patients by the first
     4                        0             0            0               1          2   0         3        reviewer and in 110 (93%) by the sec-
     6                        0             0            0               0          0   3         3        ond reviewer, the agreement was ex-
     Total                   53            45           12               3          2   3       118        cellent for both reviewers (
   C: Agreement between First and Second Reviewers                                                         0.881      0.043 for the first reviewer
                                                               Second Reviewer                             and       0.863 0.047 for the second
   First Reviewer            0            1             2            3              4   6       Total      reviewer). The interobserver agree-
     0                       53            5             1               0          0   0        59        ment for the presence of bladder le-
     1                        0           38             3               0          0   0        41        sions in each patient was noted in 111
     2                        0            2             7               0          0   0         9        (94%) of the 118 patients, thereby in-
     3                        0            0             1               3          0   0         4        dicating excellent agreement (
     4                        0            0             0               0          2   0         2        0.881 0.043).
     6                        0            0             0               0          0   3         3
     Total                   53           45            12               3          2   3       118        Diagnostic Accuracy for Lesion Detection
                                                                                                           Per lesion analysis.—The sensitivity and
   Note.—Data are numbers of patients.
                                                                                                           specificity were 92% and 97% for the

Radiology: Volume 245: Number 3—December 2007                                                                                                    801
GENITOURINARY IMAGING: Hematuria: Portal Venous Phase CT of the Bladder                                                                                    Park et al

                                                  Figure 2
first reviewer and 89% and 88% for the
second reviewer, respectively (Table 3).
Both reviewers concurred in identifying
91 (88%) of 104 lesions but missed six
lesions. There was no significant differ-
ence between the two reviewers in sen-
sitivity, specificity, positive predictive
value, negative predictive value, and
overall accuracy for detecting each blad-
der lesion (P .05).
     In the six lesions missed by both
reviewers, two were 10 mm or less in
maximum diameter and two were 15–20
mm in maximum diameter. In the re-
maining two lesions, one was masked
over the artifact by metallic material
(Fig 3), and the other, located in the            Figure 2: Transitional cell carcinoma in a 75-year-old man. (a) Contrast material– enhanced transverse
bladder base, was misinterpreted as               multi– detector row CT image shows an 11-mm hyperenhanced nodule (arrows) on the left wall of the blad-
part of the prostate.                             der. (b) The nodule (arrows) is also identified at conventional cystoscopy. Both the first and second reviewers
     In the 35 lesions with a maximum             identified this lesion.
diameter less than 10 mm, the sensitiv-
ity for lesion detection was 83% (29 of
35) for the first reviewer and 80% (28 of
35) for the second reviewer. Both re-             Discussion
                                                                                                              Table 2
viewers concurred in identifying 26             In terms of the accuracy of portal ve-
(74%) of 35 lesions (Fig 4) but missed          nous phase multi– detector row CT for                         Agreement on the Presence or
four (11%) lesions.                             bladder lesion detection, results of our                      Absence of Bladder Lesions in Each
     Per patient analysis.—The diagnos-         prospective study showed high per le-                         Patient
tic accuracy was evaluated for deter-           sion sensitivity (89%–92%) and speci-                         A: Cystoscopic Findings and First Reviewer
mining whether each patient had a               ficity (88%–97%) and per patient sensi-                                                   First Reviewer
bladder lesion (Table 3). The sensitiv-         tivity (95%) and specificity (91%–93%)                         At Cystoscopy      Absent      Present     Total
ity and specificity were 95% and 93%             for both reviewers. The per patient
                                                                                                               Absent             51            4           55
for the first reviewer and 95% and               agreement between the CT interpreta-
                                                                                                               Present             3           60           63
91% for the second reviewer, respec-            tion and the cystoscopic findings was
                                                                                                               Total              54           64          118
tively. For both reviewers, all statistical     excellent because the weighted value
                                                                                                              B: Cystoscopic Findings and Second Reviewer
parameters were 91% or greater. Both            was 0.863– 0.881 for both reviewers. In
                                                                                                                                        Second Reviewer
reviewers concurred in identifying 57           addition, the per patient negative pre-
                                                                                                              At Cystoscopy       Absent    Present     Total
(90%) of 63 patients with bladder le-           dictive value, which indicates the risk of
sions but missed two (3%) patients with         missing a disease and thereby causing                          Absent             50            5           55
bladder lesions. False-negative findings         delayed diagnosis or treatment, was                            Present             3           60           63
by both reviewers in two patients in-           94% for both reviewers. These data                             Total              53           65          118
cluded a patient with cystitis glandularis      suggest that portal venous phase multi–                       C: First and Second Reviewers
manifesting as a shallow ulcer at con-          detector row CT offers high accuracy                                                       Second Reviewer
ventional cystoscopy and a patient with         for bladder lesion detection.                                 First Reviewer      Absent       Present     Total
a 1.5-cm transitional cell carcinoma that            In terms of reader agreement for
                                                                                                               Absent             52            6           58
was obscured by a metallic artifact. One        bladder lesion detection, in our study
                                                                                                               Present             1           59           60
false-negative finding by the first re-           the agreement between the two review-
                                                                                                               Total              53           65          118
viewer in one patient was a focus of            ers with different experience level was
carcinoma in situ manifesting as color          excellent; the weighted       value was                       Note.—Data are numbers of patients.
change at conventional cystoscopy, and          0.839 in per lesion analysis and 0.881 in
one false-negative finding by the second         the per patient analysis.
reviewer in one patient was an 8-mm                  A major criticism of portal venous                   were also unsatisfactory from the per-
transitional cell carcinoma (Fig 5).            phase multi– detector row CT is its un-                   spective that the sensitivity of multi–
There was no significant difference in           deniable limitation for detecting small                   detector row CT was 80%– 83% in de-
measures of diagnostic accuracy be-             bladder lesions, as noted in previous                     tecting lesions less than 10 mm in diam-
tween the two reviewers (P .05).                studies (7–10). The results of our study                  eter.

802                                                                                                               Radiology: Volume 245: Number 3—December 2007
GENITOURINARY IMAGING: Hematuria: Portal Venous Phase CT of the Bladder                                                                                               Park et al

   There are benefits to using multi–                             upper and lower urinary tracts in one                      biopsy can improve the local staging of
detector row CT as the initial bladder                           session (14–16,19). Second, perform-                       bladder cancer because multi– detector
examination. First, multi– detector row                          ing multi– detector row CT before con-                     row CT does not induce tissue edema
CT can be used to evaluate both the                              ventional cystoscopy or transurethral                      and perivesical changes that can mimic
                                                                                                                            perivesical tumor extension.
   Table 3                                                                                                                       There are disadvantages of using
                                                                                                                            multi– detector row CT compared with
   Diagnostic Accuracy for Bladder Lesion Detection
                                                                                                                            conventional cystoscopy. First, patients
                                                                                  Reviewer                                  with impaired renal function cannot re-
   Finding                                               First                                          Second              ceive contrast material. Second, in pa-
    Per lesion analysis
                                                                                                                            tients with metal artifacts such as a me-
      True-positive                                      96                                             93
                                                                                                                            tallic prosthesis in the pelvis as noted in
      False-positive                                      2                                              7                  our study, multi– detector row CT may
      True-negative                                      58                                             53                  be limited in bladder lesion detection.
      False-negative                                      8                                             11                  Third, in patients with previous trans-
      Sensitivity (%)                                    92 (96/104)                                    89 (93/104)         urethral resection or anticancer drug in-
      Specificity (%)                                     97 (58/60)                                     88 (53/60)          stillation, procedure-related inflamma-
      PPV (%)                                            98 (96/98)                                     93 (93/100)         tion may interfere with interpretation of
      NPV (%)                                            88 (58/66)                                     83 (53/64)          CT images, thereby leading to the false-
      Accuracy (%)                                       94 (154/164)                                   89 (146/164)        positive diagnosis of lesions (7). Last, a
    Per patient analysis                                                                                                    bladder lesion adjacent to the prostate
      True-positive                                      60                                             60                  may be mistaken for the prostate, as
      False-positive                                      4                                              5                  occurred once in this series.
      True-negative                                      51                                             50                       We believe our study has a number
      False-negative                                      3                                              3                  of positive features. First, the criteria of
      Sensitivity (%)                                    95 (60/63)                                     95 (60/63)          patient enrollment in our study were
      Specificity (%)                                     93 (51/55)                                     91 (50/55)          close to the actual indications for con-
      PPV (%)                                            94 (60/64)                                     92 (60/65)
                                                                                                                            ventional cystoscopy in daily practice.
      NPV (%)                                            94 (51/54)                                     94 (50/53)
                                                                                                                            Our study included patients with pain-
      Accuracy (%)                                       94 (111/118)                                   93 (110/118)
                                                                                                                            less gross hematuria or recurrent micro-
   Note.—Data in parentheses are raw data used to calculate percentages. NPV   negative predictive value, PPV    positive   scopic hematuria, whereas most previous
   predictive value.                                                                                                        studies included patients who had al-

 Figure 3                                                         Figure 4

 Figure 3: Transitional cell carcinoma in a 69-
 year-old man. Portal venous phase transverse
 multi– detector row CT image shows a 15-mm
 hyperenhanced nodule (arrows) with calcification
 on the left wall of the bladder, which is distended              Figure 4: Transitional cell carcinoma in a 55-year-old man. (a) Portal venous phase transverse multi–
 and appears somewhat lobulated. This nodule was                  detector row CT image shows a 3-mm hyperenhanced nodule (arrow) on the posterior wall of the blad-
 not initially identified by either of two reviewers               der (b) The nodule (arrows) is also identified at conventional cystoscopy. Both the first and second reviewers
 owing to metallic artifact (arrowheads).                         identified this lesion.

Radiology: Volume 245: Number 3—December 2007                                                                                                                              803
GENITOURINARY IMAGING: Hematuria: Portal Venous Phase CT of the Bladder                                                                 Park et al

                                                                                              Figure 5
ready had bladder cancer diagnosed.              terpretation of lymph node status or
Second, because our portal venous phase          other retroperitoneal changes, our rou-
multi– detector row CT images were in-           tine CT urography protocol requires
terpreted independently by two review-           oral administration of contrast material.
ers with different levels of clinical expe-          There may be a concern about the
rience, we minimized a possible bias re-         reader agreement on the degree of con-
lated to their different levels of experience.   trast enhancement in the bladder le-
Third, our study was performed in a              sion. This study would have been more
prospective fashion, whereas most pre-           robust if readers had estimated the de-
vious studies were performed in a retro-         gree of contrast enhancement and ex-
spective fashion. Last, our study in-            tracted the threshold level for determin-
cluded 118 patients, whereas most pre-           ing the presence of bladder lesions.
vious studies included 40 or fewer               However, we did not quantify the
patients.                                        amount of the contrast enhancement in
     There are, however, limitations to          each bladder lesion for two reasons.
our study. First, although portal venous         First, it was impossible to estimate the
phase multi– detector row CT and con-            degree of contrast enhancement in
ventional cystoscopy were performed              small or flat lesions because the 2-mm
prospectively, the timing of each exami-         effective section thickness of our CT
nation was not strictly controlled be-           protocol could not obviate partial vol-      Figure 5: Transitional cell carcinoma in a 46-
cause we performed the examinations              ume averaging in lesions with a height       year-old woman. Portal venous phase transverse
according to scheduling availability—in          or a diameter less than 4 mm. Second,        multi– detector row CT image shows an 8-mm
some patients, as many as 29 days                before beginning our study, observers        hyperenhanced nodule (arrows) on the posterior
lapsed between multi– detector row CT            reached agreement on abnormal con-           wall of the bladder, which was also identified at
and cystoscopy. Second, there might              trast enhancement by means of visual         conventional cystoscopy (not shown). The first
have been a training effect for the less-        inspection in several cases.                 reviewer identified this lesion, but the second
experienced reviewer during the 7 months             There may be a criticism about a         reviewer did not.
of the study. Third, there might have            possible patient selection bias because
been a mismatch for counting the num-            only 25% (118 of 471) of patients who
ber of bladder lesions in patients with          underwent multi– detector row CT un-
multiple small lesions crowded into a            derwent cystoscopy. Our study was pro-      Acknowledgment: The authors thank Bonnie
particular focal area. Last, the scanning        spectively designed, but patients in        Hami, MA, Department of Radiology, University
delay in our study was fixed in all pa-           whom our urologist could not find a rea-     Hospitals Health System, Cleveland, Ohio, for
tients by referring to the results of a          sonable indication for conventional cys-    her editorial assistance in preparing the manu-
previous study with four-detector row                                                        script.
                                                 toscopy did not undergo it. We believe
CT (7). Because the maximum contrast             that patients who had a possibility of a
material enhancement time of a bladder           bladder lesion underwent conven-            References
lesion may vary according to various             tional cystoscopy in our study, so the       1. Sutton JM. Evaluation of hematuria in adults.
factors, such as a patient’s cardiovascu-        effect of patient selection bias was not        JAMA 1990;263:2475–2480.
lar function and the injection rate of the       severe.                                      2. Webb JA. Imaging in haematuria. Clin Ra-
contrast material, our scanning delay                In our study, we analyzed data only         diol 1997;52:167–171.
may have been suboptimal in some pa-             in patients with gross hematuria or re-
                                                                                              3. Kim JK, Ahn JH, Park T, Ahn HJ, Kim CS,
tients. Moreover, the maximum con-               current microscopic hematuria. There-           Cho KS. Virtual cystoscopy of the contrast
trast material enhancement time of cys-          fore, our results may not be fully appli-       material-filled bladder in patients with gross
titis may be different from that of blad-        cable to patients with transient or epi-        hematuria. AJR Am J Roentgenol 2002;179:
der cancer. This limitation may reduce           sodic microscopic hematuria.                    763–768.
the diagnostic performance.                          In conclusion, portal venous phase       4. Vining DJ, Zagoria RJ, Liu K, Stelts D. CT
     There may be an argument about              multi– detector row CT can provide              cystoscopy: an innovation in bladder imag-
administration of oral contrast material         high accuracy and reader agreement              ing. AJR Am J Roentgenol 1996;166:409 –
in our CT protocol because oral admin-           for bladder lesion detection in patients        410.
istration can make assessment of stones          with painless gross hematuria and re-        5. Fenlon HM, Bell TV, Ahari HK, Hussain S.
in the calyces, ureters, and bladder             current microscopic hematuria. We               Virtual cystoscopy: early clinical experience.
more difficult. Our routine CT urogra-            believe portal venous phase multi–              Radiology 1997;205:272–275.
phy protocol is aimed at evaluating all          detector row CT can be used as the           6. Radiologic investigation of patients with he-
possible disease in the urinary tract.           initial bladder examination in such pa-         maturia. Expert panel on urologic imaging.
Consequently, to achieve accurate in-            tients.                                         National Guideline Clearinghouse Web site.

804                                                                                                Radiology: Volume 245: Number 3—December 2007
GENITOURINARY IMAGING: Hematuria: Portal Venous Phase CT of the Bladder                                                                         Park et al

    http://www.guideline.gov/summary/summary        11. Kim B, Semelka RC, Ascher SM, Chalpin               Multi-detector row CT urography in the eval-
    .aspx?ss 14&doc_id 8279&string. Accessed            DB, Carroll PR, Hricak H. Bladder tumor             uation of hematuria. RadioGraphics 2003;
    February 18, 2007.                                  staging: comparison of contrast-enhanced            23:1441–1446.
 7. Kim JK, Park SY, Ahn HJ, Kim CS, Cho KS.            CT, T1- and T2-weighted MR imaging, dy-
                                                        namic gadolinium-enhanced imaging, and          16. Kawashima A, Vrtiska TJ, LeRoy AJ,
    Bladder cancer: analysis of multi-detector
                                                        late gadolinium-enhanced imaging. Radiol-           Hartman RP, McCollough CH, King BF
    row helical CT enhancement pattern and ac-
                                                        ogy 1994;193:239 –245.                              Jr. CT urography. RadioGraphics 2004;24
    curacy in tumor detection and perivesical
                                                                                                            (suppl 1):S35–S38.
    staging. Radiology 2004;231:725–731.            12. Tavernaraki AA SM, Deliveliotis C, Thanos
 8. Song JH, Francis IR, Platt JF, et al. Bladder       AI, Stasis A, Pavlopoulos SA. Virtual cystos-   17. Maher MM, Kalra MK, Rizzo S, Mueller PR,
    tumor detection at virtual cystoscopy. Radi-        copy (VC): early clinical experience [abstr].       Saini S. Multidetector CT urography in imag-
    ology 2001;218:95–100.                              Radiology 1998;209(P):592.                          ing of the urinary tract in patients with he-
                                                                                                            maturia. Korean J Radiol 2004;5:1–10.
 9. Kim JK, Park SY, Kim HS, Kim SH, Cho KS.        13. Gualdi GF, Casciani E, Rojas M, Polettini E.
    Comparison of virtual cystoscopy, multipla-         Virtual cystoscopy of bladder neoplasms:              ¨
                                                                                                        18. Gonen M, Panageas KS, Larson SM. Statisti-
    nar reformation, and source CT images with          preliminary experience [in Italian]. Radiol         cal issues in analysis of diagnostic imaging
    contrast material-filled bladder for detecting       Med (Torino) 1999;97:506 –509.                      experiments with multiple observations per
    lesions. AJR Am J Roentgenol 2005;185:
                                                    14. Caoili EM, Cohan RH, Korobkin M, et al.             patient. Radiology 2001;221:763–767.
    689 – 696.
                                                        Urinary tract abnormalities: initial experi-
10. Bernhardt TM, Rapp-Bernhardt U. Virtual                                                             19. Kim JK, Cho KS. Pictorial review: CT
                                                        ence with multi-detector row CT urography.
    cystoscopy of the bladder based on CT and                                                               urography and virtual endoscopy—promis-
                                                        Radiology 2002;222:353–360.
    MRI data. Abdom Imaging 2001;26:325–                                                                    ing imaging modalities for urinary tract
    332.                                            15. Joffe SA, Servaes S, Okon S, Horowitz M.            evaluation. Br J Radiol 2003;76:199 –209.

Radiology: Volume 245: Number 3—December 2007                                                                                                        805