Accepted Manuscript

Title: Diagnostic Performance of MRI for Prediction of

Muscle-invasiveness of Bladder Cancer: A Systematic Review
and Meta-analysis

Authors: Sungmin Woo, Chong Hyun Suh, Sang Youn Kim,

Jeong Yeon Cho, Seung Hyup Kim

PII: S0720-048X(17)30305-4
DOI: http://dx.doi.org/doi:10.1016/j.ejrad.2017.07.021
Reference: EURR 7910

To appear in: European Journal of Radiology

Received date: 27-4-2017

Revised date: 30-6-2017
Accepted date: 25-7-2017

Please cite this article as: Woo Sungmin, Suh Chong Hyun, Kim Sang Youn, Cho Jeong
Yeon, Kim Seung Hyup.Diagnostic Performance of MRI for Prediction of Muscle-
invasiveness of Bladder Cancer: A Systematic Review and Meta-analysis.European
Journal of Radiology http://dx.doi.org/10.1016/j.ejrad.2017.07.021

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
 1

Diagnostic Performance of MRI for Prediction of Muscle-invasiveness of

Bladder Cancer: A Systematic Review and Meta-analysis

Sungmin Woo, M.D. a, Chong Hyun Suh, M.D. b, c, Sang Youn Kim, M.D. a,
Jeong Yeon Cho, M.D. a, d, Seung Hyup Kim, M.D. a, d

a
Department of Radiology, Seoul National University College of Medicine, 101 Daehak-ro,
Jongno-gu, Seoul 110-744, Korea
b
Department of Radiology and Research Institute of Radiology, University of Ulsan College
of Medicine, Asan Medical Center, 86 Asanbyeongwon-Gil, Songpa-Gu, Seoul 138-736,
Republic of Korea
c
Department of Radiology, Namwon Medical Center, 365, Chungjeong-ro, Namwon-si,
Jeollabuk-do 590-702, Republic of Korea
d
Institute of Radiation Medicine and Kidney Research Institute, Seoul National University
Medical Research Center, Seoul 110-744, Korea

Manuscript type: Meta-analysis

Address correspondence to:

Sang Youn Kim, MD
Department of Radiology, Seoul National University Hospital,
101 Daehak-ro, Jongno-gu, Seoul 110-744, Korea
Tel: 82-2-2072-4897
Fax: 82-2-743-6385
E-mail: iwishluv@empas.com

Abstract

Purpose: To review the diagnostic performance of ≥1.5-Tesla MRI for local staging of bladder cancer.

Methods: MEDLINE and EMBASE were searched up to February 21, 2017. We included diagnostic accuracy

studies published since 2000 that used ≥1.5-Tesla MRI for local staging (≥T2 [muscle-invasive]) in patients with

bladder cancer, using pathology as the reference standard. The methodological quality was assessed using Quality

Assessment of Diagnostic Accuracy Studies-2. Sensitivity and specificity were pooled and plotted in a hierarchical

summary receiver operating characteristics plot. Sensitivity analyses using several clinically relevant covariates

were performed.

Results: 24 studies (1774 patients) were included. Pooled sensitivity was 0.92 (95% CI 0.88-0.95) with specificity

of 0.87 (95% CI 0.78-0.93). Sensitivity analyses showed that sensitivity estimates were comparable and
 2

consistently high across all subgroups, but specificity estimates were variable. Studies using 3-Tesla scanners had

higher specificity (0.93 [95% CI 0.86-0.98]) than those using 1.5-Tesla scanners (0.83 [95% CI 0.74-0.98]).

Studies using multiparametric MRI (conventional + ≥2 functional sequences) showed the highest accuracy with

sensitivity and specificity of 0.94 (95% CI 0.89-1.00) and 0.95 (95% CI 0.89-0.98), respectively.

Conclusions: MRI shows good diagnostic performance for predicting muscle-invasiveness of bladder cancer.

Multiparametric 3-Tesla MRI seems to improve both sensitivity and specificity.

Abbreviations:

MIBC = muscle-invasive bladder cancer

TUR = transurethral resection

MRI = magnetic resonance imaging

T1WI = T1-weighted imaging

T2WI = T2-weighted imaging

CE = contrast-enhanced

DWI = diffusion-weighted imaging

PRISMA = Preferred Reporting Items for Systematic Reviews and Meta-Analyses

QUADAS-2 = Quality Assessment of Diagnostic Accuracy Studies-2

HSROC = hierarchical summary receiver operating characteristic

Key words: Bladder cancer; Muscle-invasive bladder cancer; Local staging; Magnetic
Resonance Imaging; Systematic review; Meta-analysis

1. Introduction

In bladder cancer, determining the depth of invasion is of paramount importance. Management options differ

considerably for muscle-invasive bladder cancer (MIBC) with stage T2 or higher and non-MIBC with stage T1

or lower. MIBC is usually treated with radical cystectomy, or a combination of radiation and/or chemotherapy

while non-MIBC can be managed with transurethral resection (TUR) [1, 2]. Therefore, it would be valuable if

preoperative imaging studies can accurately assess the depth of invasion of bladder cancer.
 3

During the past two decades, there has been great interest in the use of magnetic resonance imaging (MRI) for

local staging of bladder cancer paralleling advances in MRI technique (i.e., widespread use of scanners with

magnet strength of ≥1.5 Tesla). Variable techniques including conventional T1- (T1WI) and T2-weighted imaging

(T2WI) and more advanced sequences such as contrast-enhanced (CE) MRI and diffusion-weighted imaging

(DWI) has shown promising results for determining the depth of invasion [3-5]. However, in clinical practice,

TUR followed by pathological investigation is used for local staging, and current guidelines state that due to (1)

the variability of diagnostic performance for local staging using MRI in the literature (73% to 96%) and (2)

insufficient data on the use of advanced MRI techniques (i.e., DWI), no recommendations can be made regarding

their use in the management of bladder cancer [6, 7]. In this clinical context, a comprehensive review of the

existing literature on the diagnostic performance of MRI in local staging of bladder cancer may be needed to

ascertain the role of MRI.

Therefore, we performed a systematic review and meta-analysis to evaluate the diagnostic performance of

contemporary MRI for local staging of bladder cancer with focus on the prediction of muscle-invasiveness.

2. Materials and methods

The current meta-analysis was written according to the Preferred Reporting Items for Systematic Reviews and

Meta-Analyses (PRISMA) guidelines. The research question for the purpose of this meta-analysis was formulated

based on the following PICOS criteria [8]: What is the diagnostic performance of ≥1.5-Tesla MRI for local staging

in patients with bladder cancer, as compared with pathological results in original articles published since 2000?

2.1 Literature search

A computerized search of MEDLINE and EMBASE databases up to February 21, 2017 was performed to

identify relevant studies. The search query combined synonyms for bladder cancer, MRI, staging, and diagnostic

accuracy as the following: (bladder OR urothelial OR papillary OR transitional) AND (cancer OR carcinoma OR

tumor OR tumour OR neoplasm) AND (MRI OR MR OR "magnetic resonance") AND (stage OR staging OR

TNM) AND (sensitivity OR specificity OR accuracy OR "predictive value"). References of the identified papers

were screened to expand the scope of literature search. The search was not limited to any particular language.

2.2 Study selection

2.2.1 Inclusion criteria

 4

We included studies that met the following PICOS criteria (10): (a) patients diagnosed with bladder cancer; (b)

MRI used as the index test for local staging (assessment of stage ≥T2 [MIBC]); (c) pathology based on cystectomy

or TUR for comparison; (d) sufficient information for reconstruction of 2 x 2 tables regarding sensitivity and

specificity; and (e) publication type of original articles.

2.2.2 Exclusion criteria

The exclusion criteria were as follows: (1) study population of less than 10 patients; (2) publication type other

than original articles (i.e., review articles, letters, editorials, conference abstracts, and etc.); (3) MRI used for

assessment of bladder cancer, but focusing on other topics (i.e., treatment response, local staging but not MIBC

vs non-MIBC, nodal or distant metastasis staging); (4) studies using <1.5-Tesla MRI scanners; (5) studies

published before 2000; (6) overlapping patient population; and (7) insufficient data for reconstruction of 2 x 2

tables. When multiple studies with overlapping study population were present, we only included the study with

the largest study population. Authors of the studies were contacted when 2 x 2 tables could not be reconstructed

for diagnostic accuracy studies.

The literature search and study selection process was independently performed by two reviewers and

disagreements were resolved by consensus.

2.3. Data extraction and quality assessment

The following data with regards to patient/tumor, study, and MRI characteristics were extracted from the

selected studies by using a standardized form:

(1) Patient and tumor characteristics―number of patients, ethnicity of study population, mean age and range

of patients, number of tumors, prevalence of stage ≥T2 tumors, and histological subtypes (only urothelial

carcinoma vs inclusion of other subtypes)

(2) Study characteristics―origin of study (authors, institution, and duration of patient recruitment), publication

year, study design (i.e., prospective, multi-center, and consecutive enrollment), reference standard (based on

cystectomy or TUR), interval between MRI and reference standard, blinding between MRI interpretation and

pathological assessment, and characteristics of readers (number, consensus reading, and experience)

(3) MRI characteristics―scanner manufacturer and model, magnet field strength (1.5- or 3.0-Tesla), type of

MRI sequences used, their corresponding technical parameters, whether preparation for bladder distension was

performed or not, whether MRI was performed before or after a prior TUR or biopsy, and whether criteria for

staging was explicitly provided.

 5

The methodological quality of the included studies was assessed using the Quality Assessment of Diagnostic

Accuracy Studies-2 (QUADAS-2) tool [9]. Data extraction and quality assessment was performed independently

by two reviewers with disagreements resolved via consensus.

2.4 Data synthesis and analysis

Two x two tables were reconstructed from the included studies to calculate their sensitivity and specificity. If

several different diagnostic performance values were provided within a single study (i.e., several types of MRI

sequences or multiple independent readers), we selected the result with the highest accuracy to be representative

of the study.

Summary estimates of sensitivity and specificity were calculated with hierarchical logistic regression modelling

including bivariate and hierarchical summary receiver operating characteristic (HSROC) modelling [10, 11].

These results were plotted using HSROC curves with 95 % confidence and prediction regions. Publication bias

was evaluated using visual analysis of the Deeks’ funnel plot and calculating the p value using Deeks’ asymmetry

test [12].

Heterogeneity was assessed with the following methods: (1) Cochran’s Q-test with p <0.05 indicating that

heterogeneity was present; (2) Higgins I2 test with the degree of heterogeneity interpreted using the following

criteria: inconsistency index (I2) = 0%–40%, heterogeneity might not be important; 30%–60%, moderate

heterogeneity may be present; 50%–90%, substantial heterogeneity may be present; and 75%–100%, considerable

heterogeneity [13]; and (3) testing for a threshold effect (positive correlation between sensitivity and false positive

rate) between the included studies. In order to explore the cause of heterogeneity, sensitivity analyses using several

covariates were performed.

The “metandi” and “midas” modules in Stata 10.0 (StataCorp LP, College Station, TX, USA) and “mada”

package in R software version 3.2.1 (R Foundation for Statistical Computing, Vienna, Austria) were used for

statistical analyses with p <0.05 indicating statistical significance.

3. Results

3.1. Literature search

The systematic literature search initially yielded 1062 articles. After removing 387 duplicates, screening of the

675 titles and abstracts yielded 70 potentially eligible original articles. Full-text reviews were performed and 46

studies were excluded due to the following: less than 10 patients (n = 1), not in field of interest (n = 4), used <1.5-
 6

Tesla MRI (n = 24), published before 2000 (n = 11), insufficient data to reconstruct 2 x 2 tables (n = 3), and shared

study population with other studies (n = 3). Ultimately, 24 studies including 1774 patients evaluating the

diagnostic performance of MRI for local staging of bladder cancer were included in this meta-analysis [3-5, 14-

34]. The study selection process is shown in Figure 1.

3.2. Characteristics of included studies

The patient characteristics are summarized in Table 1. The size of the study population ranged from 19 to 362

patients. The patients were of Asian ethnicity in 13 studies and non-Asian in the other 11. The mean age of the

patients ranged from 61.3 to 72.1 years. The number of bladder cancers ranged from 20 to 362. The prevalence

of stage ≥T2 tumors ranged from 14% to 96%. 10 studies only included patients with urothelial carcinoma, while

other histological subtypes were also included in 10 and details were not provided in four.

The study characteristics are described in Table 2. The study design was prospective in 10 studies, and

retrospective in 14. All but two studies were single-center studies. Patient recruitment was consecutive in all but

two studies. Four studies used only cystectomy as the reference standard; three used only TUR; and 16 used either

cystectomy or TUR. The interval between MRI and the reference standard was provided in 11 studies. MRI was

interpreted blinded to the reference standard in most (17 of 24) of the studies.

The MRI characteristics are shown in Table 3. Six studies used 3-Tesla scanners, while 18 used 1.5-Tesla

scanners. Various sequences were used in the included studies. Three studies used only conventional sequences

(T1WI and/or T2WI); six used only a single functional imaging technique; 10 used one functional imaging

technique in addition to conventional sequences; and five used two or more functional imaging techniques in

addition to conventional sequences. Bladder distension consisting of a combination of methods including

prohibiting the patients from urination, oral hydration, or saline injection via Foley catheter was performed in 11

studies. MRI was performed before TUR or biopsy in all patients in 14 studies, in some in three studies, and after

TUR or biopsy in five. The criteria for local staging on MRI was explicit in most (20 of 24) studies.

3.3. Quality assessment

In general, the quality of the studies was considered moderate, with 21 of the 24 studies satisfying at least four

of the seven QUADAS-2 domains (Figure 2). Regarding the patient selection domain, two studies were considered

to have high risk of bias as they excluded patients with stage T3-4 bladder cancer when differentiating between

≤T1 and ≥T2 tumors [5, 24]. Another two studies were considered to have unclear risk of bias as they did not

explicitly mention whether patient enrollment was consecutive or not [22, 29]. Regarding the index test domain,
 7

there was high risk of bias in two studies as the threshold used for determining local staging was not pre-specified

[3, 33]. There was unclear risk of bias in 10 studies, as blinding was unclear in seven studies [15-18, 22, 29, 31]

and it was unclear whether a pre-specified threshold was used in three studies [20, 28, 34]. Regarding the reference

standard domain, all but five [4, 20, 25, 31] of the studies had high risk of bias as the reference standard was based

on TUR is all or some of the patients. However, those five studies had unclear risk of bias, as it was only described

that pathological results were used and detailed methods including the use of TUR or cystectomy was not

mentioned in one study [31] and it was unclear whether the pathology was assessed blinded to MRI in four studies

[4, 20, 22, 25]. Regarding the flow and timing domain, 16 studies had high risk of bias as different reference

standards were applied within the study [3, 5, 14-17, 19, 23, 24, 26, 28-30, 32-34]. Five studies were considered

to have unclear risk of bias as the MRI-reference standard interval was not provided [4, 20, 22, 25, 31]. There was

low concern for applicability with regard to the first three QUADAS-2 domains for all but one study [31], in

which there was unclear concern as the details for deriving the reference standard (TUR or cystectomy) was not

explicit.

3.4. Diagnostic accuracy and heterogeneity assessment

The pooled sensitivity and specificity for diagnosing MIBC were 0.92 (95% CI 0.88-0.95) and 0.87 (95% CI

0.78-0.93), respectively. The area under the HSROC curve was 0.95 (95 % CI 0.93–0.97) (Figure 3). The

Cochran’s Q-test suggested that heterogeneity was present (Q = 74.049, p <0.001). Higgins I 2 statistics

demonstrated substantial and considerable heterogeneity with regard to the sensitivity (I 2 = 70.65%) and

specificity (I2 = 90.24%), respectively. No threshold effect was shown upon visualization of the coupled forest

plot of sensitivity and specificity (Figure 4) with a correlation coefficient between sensitivity and false positive

rate of 0.270 (95% CI -0.150-0.607). The Deeks’ funnel plot suggested that publication bias was present, with a

p value of 0.01 for slope coefficient (Figure 5).

3.5. Sensitivity analysis

Figure 6 shows the results of sensitivity analyses. In general, sensitivity estimates were comparable and high

while there was notable variation in the specificity estimates. Studies assessing only urothelial carcinoma (n = 10)

showed lower specificity (0.79 [95% CI 0.65-0.98]) than those evaluating urothelial carcinoma and other tumors

(n = 10; 0.91 [95% CI 0.83-0.98]). Studies using only cystectomy as the reference standard (n = 4) showed lower

specificity (0.45 [95% CI 0.11-0.98]) than those using TUR only (n = 3; 0.84 [95% CI 0.61-0.98]) or both (n =

16; 0.91 [95% CI 0.86-0.98]). Studies using 3-Tesla scanners (n = 6) had higher specificity (0.93 [95 % CI 0.86-
 8

0.98]) than those using 1.5-Tesla scanners (n = 18; 0.83 [95% CI 0.74-0.98]). There was a trend for higher

sensitivity (0.94 [95% CI 0.89-1.00]) and specificity (0.95 [95% CI 0.89-0.98]) in studies (n = 5) using

multiparametric approach (conventional sequence + ≥2 functional techniques) than those using only conventional

sequence (n = 3; 0.86 [95% CI 0.73-0.99] and 0.83 [95% CI 0.60-0.98], respectively) or a single functional

technique (n = 6; 0.89 [95% CI 0.80-0.97] and 0.81 [95% CI 0.62-0.98], respectively). Two studies [4, 22] were

considered to be outliers showing peculiarly low specificity (0.00 and 0.29) compared with others (0.76-1.00).

Pooling the results of 22 studies excluding these two studies resulted in a sensitivity and specificity of 0.91 (95%

CI 0.87-0.94) and 0.89 (95% CI 0.84-0.93), respectively, with a substantial decrease in the degree of heterogeneity

(I2) from 70.65% to 65.66% and from 90.24% to 75.91% for sensitivity and specificity, respectively.

4. Discussion

In this meta-analysis, we evaluated the diagnostic performance of ≥1.5-Tesla MRI for local staging in bladder.

To the best of our knowledge, this is the only meta-analysis on this subject other than the study by Zhai et al [35],

which evaluated only eight studies using only DWI without any exploration of heterogeneity. The current meta-

analysis provides a comprehensive overview of the literature published since 2000, encompassing a large number

of studies (n = 24) using various MRI techniques, and provides the results of heterogeneity exploration using

extensive sensitivity analysis. Our results show that the pooled sensitivity and specificity of the 24 included studies

for differentiating MIBC from non-MIBC were 0.92 (95% CI 0.88-0.95) and 0.87 (95% CI 0.78-0.93),

respectively. The differentiation between MIBC and non-MIBC is crucial, as they show different prognosis and

generally are treated with different approaches. Therefore, it may be suggested based on the results of our meta-

analysis that MRI be incorporated in guidelines for bladder cancer as one of the primary modalities for local

staging.

There was significant heterogeneity among the included studies. However, extensive sensitivity analysis

provided some insight regarding the potential causes of heterogeneity. In general, the sensitivity estimates were

relatively comparable while variation was more notable with regard to the specificity estimates. Specifically,

studies that only evaluated urothelial carcinomas and those that used only cystectomy as the reference standard

showed lower sensitivity that other subgroups. We speculate that this was attributed to the studies by Daneshmand

et al [4] and Nishimura et al [22], which showed peculiarly lower specificity (0.29 and 0.00, respectively) than

the other 22 studies (0.76-1.00), rather than reflecting a true association with these covariates (pathological
 9

subtype and reference standard). In fact, when pooling the results of 22 studies excluding these two studies, there

was a remarkable decrease in the degree of heterogeneity. We cannot fully understand why these two studies

showed such different results from the other included studies. However, the study by Nishimura et al [22] was

based on a study population predominantly (18/27) consisting of chemotherapy-treated bladder cancer patients.

We speculate that this may have one of the major reasons why the study suffered from peculiarly low specificity.

Furthermore, although we performed sensitivity analysis for each subgroup regarding reference standard

(cystectomy only, TUR only, cystectomy or TUR), the 95% confidence intervals for summary estimates were

relatively wide as there were only a few number studies included in the cystectomy only group (n = 4) and TUR

only group (n = 3). Therefore, firm conclusions cannot be made regarding reference standard as a source of

heterogeneity and further studies are needed.

The current study highlights that the MRI protocol may largely influence the diagnostic accuracy of local

staging. Using a multiparametric approach, consisting of 2 or more functional techniques in addition to

conventional sequences (T2WI), resulted in higher accuracy compared with using only conventional sequences

or only a single functional technique. T2WI is able to differentiate the layers of the bladder as it can be acquired

with high spatial resolution. On the other hand, functional techniques also have strengths in that they can better

depict the tumor itself (DWI), or provide other imaging features that aid in the local staging such as tumor stalk

(DWI) and submucosal linear enhancement (CE-MRI) [5, 30]. Therefore, by comprehensively utilizing both the

anatomical information and functional information from T2WI and DWI or CE-MRI, the performance of MRI for

predicting muscle-invasiveness of bladder cancer may be improved. Furthermore, studies using 3-Tesla MRI

showed better specificity than those using 1.5-Tesla MRI. Therefore, it may be suggested that guidelines

recommend multiparametric 3-Tesla MRI in order to improve the diagnostic accuracy of local staging of bladder

cancer.

It is well known that optimal distension of the bladder is needed when using computed tomography or MRI for

local staging of bladder cancer. In the current meta-analysis, there was no significant differences in the accuracy

between studies that performed bladder distension and the others. However, robust conclusions cannot be made

for several reasons: (1) there were no studies that explicitly stated that they did not perform bladder distension;

(2) studies that did perform them used various methods including prohibiting patients from voiding, oral hydration,

saline infusion via Foley catheter or a combination of the above; and (3) we are only aware that those studies did
 10

perform such measures, but do not know whether the bladders were “optimally distended” or not. Further studies

may be needed to clarify this issue.

As several of the included studies provided multiple diagnostic performance values from multiple readers and

MRI protocols within the identical study population, we had to choose one to represent each study. For this meta-

analysis, we included the results with the highest accuracy, which generally corresponded to the results from the

more experienced reader and advanced MRI technique. As it is generally known the diagnostic accuracy of pelvic

MRI interpretations are more accurate by more experienced readers, there may be concern that the pooled

estimates from our study may not be directly applicable to real life clinical practice [36, 37]. However, the

differences between multiple diagnostic performance values were mostly comparable between readers―therefore,

had we selected the results from the less experienced readers there would only have been minor changes in the

results without significantly affecting the overall conclusion of our meta-analysis.

One of the major limitations of our meta-analysis is that most of the studies used either TUR or cystectomy (n

= 16) or only TUR (n = 3) as the reference standard. However, as compared with cystectomy, TUR may be less

accurate in determining the muscle-invasiveness of bladder cancer, therefore resulting in a less robust reference

standard. In order to deal with this issue, additional measures (i.e., secondary TUR or additional deep muscle

biopsy) were taken in 9 of the studies that used TUR. Another limitation of this meta-analysis is that we only

assessed the diagnostic performance for differentiating MIBC and non-MIBC but did not perform a match

between MRI and pathology for each T stage. However, we decided that determination of muscle-invasiveness of

bladder cancer would be clinical relevant as this is one of the most important factors regarding prognosis and

treatment decision. In addition, publication bias was present among the included studies in this meta-analysis. We

believe that this bias may have been present in our meta-analysis as studies with more significant results are more

likely to be published in the literature. Therefore, there may be a possibility that the diagnostic performance of

MRI in this meta-analysis for local staging of bladder cancer may be overestimated.

5. Conclusions

MRI shows good diagnostic performance for prediction of muscle-invasiveness of bladder cancer. Using

multiparametric 3-Tesla MRI seemed to improve both sensitivity and specificity.

 11

References

1. Josephson D, Pasin E, Stein JP. Superficial bladder cancer: part 2. Management. Expert Rev. Anticancer Ther.

7 (2007) 567-581.

2. Hautmann RE, Abol-Enein H, Hafez K, et al. Urinary diversion. Urology. 69 (2007) 17-49.

3. Rosenkrantz AB, Haghighi M, Horn J, et al. Utility of quantitative MRI metrics for assessment of stage and

grade of urothelial carcinoma of the bladder: preliminary results. Am. J. Roentgenol. 201 (2013) 1254-1259.

4. Daneshmand S, Ahmadi H, Huynh LN, Dobos N. Preoperative staging of invasive bladder cancer with dynamic

gadolinium-enhanced magnetic resonance imaging: results from a prospective study. Urology. 80 (2012) 1313-

1318.

5. Wang HJ, Pui MH, Guan J, et al. Comparison of Early Submucosal Enhancement and Tumor Stalk in Staging

Bladder Urothelial Carcinoma. Am. J. Roentgenol. (2016) Aug 1-7 [Epub ahead of print].

6. Babjuk M, Bohle A, Burger M, et al. EAU Guidelines on Non-Muscle-invasive Urothelial Carcinoma of the

Bladder: Update 2016. Eur. Urol. 71 (2017) 447-461.

7. Alfred Witjes J, Lebret T, Comperat EM, et al. Updated 2016 EAU Guidelines on Muscle-invasive and

Metastatic Bladder Cancer. Eur. Urol. 71 (2017) 462-475.

8. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for reporting systematic reviews and meta-

analyses of studies that evaluate health care interventions: explanation and elaboration. J. Clin. Epidemiol. 62

(2009) e1-34.

9. Whiting PF, Rutjes AW, Westwood ME, et al. QUADAS-2: a revised tool for the quality assessment of

diagnostic accuracy studies. Ann. Int. Med. 155 (2011) 529-536.

10. Suh CH, Park SH. Successful Publication of Systematic Review and Meta-Analysis of Studies Evaluating

Diagnostic Test Accuracy. Korean J. Radiol. 17 (2016) 5-6.

11. Lee J, Kim KW, Choi SH, Huh J, Park SH. Systematic Review and Meta-Analysis of Studies Evaluating

Diagnostic Test Accuracy: A Practical Review for Clinical Researchers-Part II. Statistical Methods of Meta-

Analysis. Korean J. Radiol. 16 (2015) 1188-1196.

12. Deeks JJ, Macaskill P, Irwig L. The performance of tests of publication bias and other sample size effects in

systematic reviews of diagnostic test accuracy was assessed. J. Clin. Epidemiol. 58 (2005) 882-893.

13. Higgins J, Green S. Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0. The Cochrane

Collaboration. http://handbook.cochrane.org/chapter_9/9_5_2_identifying_and_measuring_heterogeneity.htm.
 12

Updated March 2011. Accessed January 3, 2017.

14. Abdel-Rahman HM, El Fiki IM, Desoky EA, Elsayed ER, Samad KMA. The role of diffusion-weighted

magnetic resonance imaging in T staging and grading of urinary bladder cancer. Egyptian J. Radiol. Nucl. Med.

46 (2015) 741-747.

15. Afifi AH, Maksoud TASA, EL-noueam KI, Ataa MA, Abdallah DM. Multiparametric-MRI as a

comprehensive study in evaluation, characterization & local staging of urinary bladder carcinomas. Egyptian J.

Radiol. Nucl. Med. (2017) Feb 1 [Epub ahead of print].

16. Ghafoori M, Shakiba M, Ghiasi A, Asvadi N, Hosseini K, Alavi M. Value of MRI in local staging of bladder

cancer. Urol. J. 10 (2013) 866-872.

17. Hayashi N, Tochigi H, Shiraishi T, Takeda K, Kawamura J. A new staging criterion for bladder carcinoma

using gadolinium-enhanced magnetic resonance imaging with an endorectal surface coil: a comparison with

ultrasonography. BJU Int. 85 (2000) 32-36.

18. Kobayashi S, Koga F, Yoshida S, et al. Diagnostic performance of diffusion-weighted magnetic resonance

imaging in bladder cancer: potential utility of apparent diffusion coefficient values as a biomarker to predict

clinical aggressiveness. Eur. Radiol. 21 (2011) 2178-2186.

19. Lee M, Shin SJ, Oh YT, et al. Non-contrast magnetic resonance imaging for bladder cancer: fused high b

value diffusion-weighted imaging and T2-weighted imaging helps evaluate depth of invasion. Eur. Radiol. (2017)

Feb 6 [Epub ahead of print].

20. Lista F, Andres G, Caceres F, Ramon de Fata F, Rodriguez-Barbero JM, Angulo JC. Evaluation of risk of

muscle invasion, perivesical and/or lymph node affectation by diffusion-weighted magnetic nuclear resonance in

the patient who is a candidate for radical cystectomy. Actas. Urol. Esp. 37 (2013) 419-424.

21. Mizuno K, Sasaki T, Saito Y, et al. [Gadolinium-enhanced MR imaging, T2-weighted MR imaging, and

transurethral ultrasonography]. Nippon Acta. Radiol. 61 (2001) 496-501.

22. Nishimura K, Fujiyama C, Nakashima K, Satoh Y, Tokuda Y, Uozumi J. The effects of neoadjuvant

chemotherapy and chemo-radiation therapy on MRI staging in invasive bladder cancer: comparative study based

on the pathological examination of whole layer bladder wall. Int. Urol. Nephrol. 41 (2009) 869-875.

23. Ohgiya Y, Suyama J, Sai S, et al. Preoperative T staging of urinary bladder cancer: efficacy of stalk detection

and diagnostic performance of diffusion-weighted imaging at 3T. Magn. Reson. Med. Sci. 13 (2014) 175-181.

24. Panebianco V, De Berardinis E, Barchetti G, et al. An evaluation of morphological and functional multi-
 13

parametric MRI sequences in classifying non-muscle and muscle invasive bladder cancer. Eur. Radiol. (2017)

Feb 8 [Epub ahead of print].

25. Rabie E, Faeghi F, Izadpanahi MH, Dayani MA. Role of Dynamic Contrast-Enhanced Magnetic Resonance

Imaging in Staging of Bladder Cancer. J. Clin. Diagn. Res. 10 (2016) TC 01-05.

26. Rajesh A, Sokhi H, Fung R, Mulcahy K, Bankart M. Bladder cancer: evaluation of staging accuracy using

dynamic MRI. Clin. Radiol. 66 (2011) 1140-1145.

27. Sevcenco S, Ponhold L, Heinz-Peer G, et al. Prospective evaluation of diffusion-weighted MRI of the bladder

as a biomarker for prediction of bladder cancer aggressiveness. Urol. Oncol. 32 (2014) 1166-1171.

28. Sherif MF. The value of diffusion weighted MR imaging in T staging and correlation with histologic grading

in urinary bladder cancer. Egyptian J. Radiol. Nucl. Med. 46 (2015) 189-194.

29. Stojovska-Jovanovska E, Mitreska N, Stojovski M, Lazarova A, Stavridis S, Dodevski A. Computed

tomography or magnetic resonance imaging - our experiences in determining preoperative TNM staging of

bladder cancer. Pril (Makedon Akad Nauk Umet Odd Med Nauki). 34 (2013) 63-70.

30. Takeuchi M, Sasaki S, Ito M, et al. Urinary bladder cancer: diffusion-weighted MR imaging--accuracy for

diagnosing T stage and estimating histologic grade. Radiology. 251 (2009) 112-121.

31. Tillou X, Grardel E, Fourmarier M, et al. [Can MRI be used to distinguish between superficial and invasive

transitional cell bladder cancer?]. Prog. Urol. 18 (2008) 440-444.

32. Watanabe H, Kanematsu M, Kondo H, et al. Preoperative T staging of urinary bladder cancer: does diffusion-

weighted MRI have supplementary value? Am. J. Roentgenol. 192 (2009) 1361-1366.

33. Wu LM, Chen XX, Xu JR, et al. Clinical value of T2-weighted imaging combined with diffusion-weighted

imaging in preoperative T staging of urinary bladder cancer: a large-scale, multiobserver prospective study on

3.0-T MRI. Acad. Radiol. 20 (2013) 939-946.

34. Yamada Y, Kobayashi S, Isoshima S, Arima K, Sakuma H, Sugimura Y. The usefulness of diffusion-weighted

magnetic resonance imaging in bladder cancer staging and functional analysis. J. Cancer Res. Ther. 10 (2014)

878-882.

35. Zhai N, Wang YH, Zhu LM, et al. Sensitivity and Specificity of Diffusion-Weighted Magnetic Resonance

Imaging in Diagnosis of Bladder Cancers. Clin. Invest. Med. 38 (2015) E173-184.

36. Woo S, Kim SY, Cho JY, Kim SH. Assessment of deep myometrial invasion of endometrial cancer on MRI:

added value of second-opinion interpretations by radiologists subspecialized in gynaecologic oncology. Eur.

 14

Radiol. 27 (2017) 1877-1882.

37. Wibmer A, Vargas HA, Donahue TF, et al. Diagnosis of Extracapsular Extension of Prostate Cancer on

Prostate MRI: Impact of Second-Opinion Readings by Subspecialized Genitourinary Oncologic Radiologists. Am.

J. Roentgenol. 205 (2015) W73-78.

 15

Figure legends

Fig. 1. Flow diagram showing study selection process for meta-analysis.

Fig. 2. Grouped bar charts show risk of bias (left) and concerns for applicability (right) of 24 included studies

using QUADAS-2.

Fig. 3. Hierarchical summary receiver operating characteristic curve of the diagnostic performance of MRI for

diagnosis of ≥T2 bladder cancer.

Fig. 4. Coupled forest plots of pooled sensitivity and specificity. Numbers are pooled estimates with 95%

confidence intervals (CI) in parentheses. Corresponding heterogeneity statistics are provided at bottom right

corners. Horizontal lines indicate 95% CIs.

Fig. 5. Deeks’ funnel plot for per-patient analysis. P-value of 0.01 suggests that publication bias is present.

Fig . 6. Forest plots of sensitivity analyses showing pooled sensitivity and specificity estimates and

corresponding 95% confidence intervals (CIs) for various subgroups. TUR = transurethral resection; * = unstable

pooled estimates
 16

Figure 1

Figure 2
 17

Figure 3
 18

Figure 4
 19

Figure 5

Figure 6
 20

Table 1. Patient and tumor characteristics

Patient characteristics Tumor characteristics Raw data for analysis
Author No. Age (yrs) No. Pathological T stage Details of subtypes (n) TP FN FP TN
Ethnicity Histological subtype
(n) Mean Range (n) Range ≥T2 (%) TCC SqCC Adca other
Abdel-Rahman HM [14] 40 Non-Asian NR 52-70 40 ≤T1-T4 65 Urothelial carcinoma only 34 4 2 22 4 2 12
Afifi AH [15] 50 Non-Asian 59.8 30-82 NR T1-T4 36 Other subtypes included 48 1 1 14 4 2 30
Daneshmand S [4] 122 Non-Asian 67.8 46-81 122 T0-T4 66 Urothelial carcinoma only 122 72 8 30 12
Ghafoori M [16] 86 Asian 59.7 32-86 108 Ta-T4 90 NR 95 2 2 9
Hayashi N [17] 71 Asian 65.5 31-85 71 ≤T1-T4 15 NR 10 1 8 52
Kobayashi S [18] 104 Asian 68 a 38-88 104 Ta-≥T2 37 Other subtypes included 98 5 1 26 12 6 60
a
Lee M [19] 62 Asian 67.5 41-92 NR <T2-≥T2 42 Urothelial carcinoma only 62 21 5 8 28
Lista F [20] 20 Non-Asian 69.0 NR 20 T0-T4 65 Urothelial carcinoma only 20 13 0 1 6
Mizuno K [21] 64 Asian 68.2 16-89 64 T1-T4 42 Urothelial carcinoma only 64 25 2 9 28
Nishimura K [22] 27 Asian 67.5 NR 27 T2-T3 78 Other subtypes included 20 5 1 1 21 0 6 0
Ohgiya Y [23] 39 Asian 70.7 43-90 NR T1-T4 38 Other subtypes included 38 1 13 2 4 20
Panebianco V [24] 76 Non-Asian 66.0 45-80 82 Ta-T4 35 NR 27 2 3 50
Rabie E [25] 45 Asian 52.1 43-63 45 Ta-T4 96 Urothelial carcinoma only 45 42 1 0 2
Rajesh A [26] 100 Non-Asian NR 55-95 100 Ta-T4 45 Urothelial carcinoma only 100 42 3 12 43
Rosenkrantz AB [3] 37 Non-Asian NR NR 37 Ta-T4 43 Urothelial carcinoma only 37 15 1 5 16
Sevcenco S [27] 49 Non-Asian 68 a 30-90 49 Ta-≥T2 20 Other subtypes included 43 6 9 1 7 26
Sherif MF [28] 30 Non-Asian 65.0 50-80 30 T1-T4 80 Other subtypes included 28 2 24 0 0 6
Stojovska-Jovanovska E [29] 90 Non-Asian NR 34-85 90 Ta-T4 47 NR 33 9 4 44
Takeuchi M [30] 40 Asian 70 a 49-85 52 Tis-T4 33 Other subtypes included 50 1 1 16 1 0 35
Tillou X [31] 60 Non-Asian 66.8 44-91 60 Ta-T4 17 Urothelial carcinoma only 60 9 1 10 40
Wang HJ [5] 59 Asian 61.3 31-82 92 ≤T1-T2 14 Urothelial carcinoma only 92 9 4 4 75
Watanabe H [32] 19 Asian 71.0 55-83 NR T1-T4 28 Other subtypes included 18 1 4 1 3 11
Wu LM [33] 362 Asian 71 48-87 362 Tis-T4 29 Other subtypes included 343 16 3 99 6 0 257
Yamada Y [34] 123 Asian 72.1 38-90 NR Tis-T4 46 Urothelial carcinoma only 123 54 2 15 57
Adca = adenocarcinoma; FN = false negative; FP = false positive; N/A = not available; NR = not reported; SqCC = squamous cell carcinoma; TN = true negative; TP = true positive
a
median
 21

Table 2. Study characteristics

Origin Design Reference standard Reader characteristics
Publication
Author Consecutive MRI-reference Reader
year Period Institution Prospective Multicenter Method Number Consensus Blinded
enrollment standard interval experience (y)
Cystectomy or
Abdel-Rahman HM [14] 2015 2012.10-2014.4 Zagazig University Yes No Yes NR 2 NR 10/1 mo Yes
TUR
Cystectomy or
Afifi AH [15] 2017 2015.6-2016.8 Alexandria University Yes No Yes 2-4wks NR NR NR NR
TUR
Cystectomy Yes
Daneshmand S [4] 2012 2005-2009 Oregon Helath and Science University Yes No Yes NR 2 No NR
only
Cystectomy or
Ghafoori M [16] 2013 2009.12-2011.4 Hazrat Rasoul Akram University Hospital No No Yes NR 1 N/A NR NR
TUR
Cystectomy or
Hayashi N [17] 2000 NR Mie University School of Medicine No No Yes NR NR NR NR NR
TUR a
b
Kobayashi S [18] 2011 2007.2-2010.5 Tokyo Medical and Dental University Graduate School Yes No Yes TUR only 2-68 NR NR NR NR
Cystectomy or Yes
Lee M [19] 2017 NR Yonsei University College of Medicine No No Yes 1-18 1 N/A 5
TUR a
Cystectomy Yes
Lista F [20] 2013 NR Hospital Universitario de Getafe Yes No Yes NR Multiple NR NR
only
Mizuno K [21] 2001 1997.2-2000.5 Hirosaki University School of Medicine No No Yes TUR only 2-65 3 No NR Yes

Cystectomy
Nishimura K [22] 2009 NR Saga University No No NR NR NR NR NR NR
only
Cystectomy or Yes Yes
Ohgiya Y [23] 2014 2007.6-2009.9 Showa University School of Medicine No No Yes 0-41 2 10/10
TUR a
Cystectomy or Yes Yes
Panebianco V [24] 2017 2016.1-2016.11 Sapienza University of Rome No No Yes 2-8wks 2 10/3
TUR
Shahid Beheshti University of Medical Sciences, Al-Zahra Cystectomy Yes
Rabie E [25] 2016 2013.1-2014.7 No Yes Yes NR 1 N/A NR
Hospital, Shahrekord University of Medical Sciences only
Cystectomy or Yes
Rajesh A [26] 2011 NR Leicester General Hospital No No Yes NR 1 N/A 5
TUR
Cystectomy or Yes
Rosenkrantz AB [3] 2013 NR NYU Langone Medical Center, Mount Sinai School of Medicine No Yes Yes 1-84 2 Yes NR/NR
TUR
Sevcenco S [27] 2014 2012-2013 Medical University of Vienna Yes No Yes TUR only 1-58 2 NR 8/6 Yes

Cystectomy or Yes
Sherif MF [28] 2015 2012.2-2014.4 Tanta University Yes No Yes NR 1 NR NR
TUR
Cystectomy or
Stojovska-Jovanovska E [29] 2013 NR University Radiology Clinic (Skopje, Republic of Macedonia) No No NR NR NR NR NR NR
TUR a
Cystectomy or No
Takeuchi M [30] 2009 2006.7-2007.11 Nagoya City University Graduate School of Medical Sciences Yes No Yes NR 2 21/15 Yes
TUR a
Tillou X [31] 2008 2002.5-2005.11 CHU Sud Yes No Yes NR NR 2 No NR NR
Cystectomy or
Wang HJ [5] 2016 2011.11-2014.12 First Affiliated Hospital of Sun Yat-sen University Yes No Yes <2wks 2 No 29/5 Yes
TUR a
Cystectomy or No Yes
Watanabe H [32] 2009 2007-2008 Gifu University Hospital No No Yes 6-30 2 NR
TUR
Cystectomy or No Yes
Wu LM [33] 2013 2007.11-2012.5 Ren Ji Hospital Yes No Yes <29 3 10/4/2
TUR a
Cystectomy or
Yamada Y [34] 2014 2006.2-2013.1 Mie University School of Medicine No No Yes NR 1 N/A NR Yes
TUR b
N/A = not available; NR = not reported; TUR = transurethral resection
a
With deep muscle biopsy
b
With confirmatory repeat TUR
 22

Table 3. MRI characteristics

Scanner Technical parameters
Magne Conventional Pre-
DWI CE-MRI Bladder Explici
t sequences TUR/
Author Manufactur Sequences distensi t
Model Strengt Biopsy
er used ST/Interval ST/Interval B-value ST/Interval Scan delay on criteria
h Plane Plane Plane MRI
(mm) (mm) (s/mm2) (mm) (sec)
(Tesla)
Abdel-Rahman HM
Philips Achieva 1.5 DWI A,S 3/1 0,500,1000 Yes Yes Yes
[14]
T1/T2/DWI/
Afifi AH [15] Siemens Magnetom Avanto 1.5 A 3/1 A 4/0.4 0,500,1000 A 6/1 20/70/180 Yes Yes Yes
CE
Daneshmand S [4] GE Signa 1.5 T1/T2/CE A,C,S 6/2 A 6/2 20/52-86 NR No Yes

Ghafoori M [16] Siemens Avanto 1.5 T1/T2/CE A,C,S NR A NR NR NR Yes Yes

Hayashi N [17] GE Signa 1.5 T1/T2/CE A,C,S NR A,C,S NR 300 NR No Yes

Kobayashi S [18] Philips Achieva 1.5 T2 A 5/0.5 NR Yes Yes

Discovery MR750,
Lee M [19] GE 3 T2/DWI A 6/NR A 6 50,1000 Yes NR Yes
MR750w
Lista F [20] Siemens Avanto 1.5 DWI A,C,S 5/NR NR NR No No

Mizuno K [21] Siemens Magnetom Vision 1.5 T2 or CE* P 4/0.2 P 4/0.2 19 Yes Yes Yes

Nishimura K [22] GE Signa 1.5 CE C NR 30 Yes Yes No

Ohgiya Y [23] Siemens Magnetom Trio 3 T2/DWI A,S 4/0.6 A 3/0.6 0,500,1000 NR Yes Yes
T2/DWI/DT 0,500,800,
Panebianco V [24] GE Discovery MR750 3 A,S 3/NR A 3/NR NR 3/NR 20-180 Yes Some Yes
I/CE 1000
T1/T2/STIR
Rabie E [25] Siemens Magnetom Avanto 1.5 A,C,S 4/1 A 3.6/0.9 NR NR No Yes
/CE
Magnetom
Rajesh A [26] Siemens 1.5 T2/CE A,C,S 5/1 A 3/0.6 20/70 NR Some Yes
Symphony
Rosenkrantz AB [3] Siemens Magnetom Avanto 1.5 T2 A,S 4/NR NR Yes Yes
50,400,100
Sevcenco S [27] Siemens Tim Trio 3 DWI A 5/NR NR Yes Yes
0
T1/T2/DWI/
Sherif MF [28] GE NR 1.5 A,C,S 3/NR A,S 3/NR 0,1000 A 3/NR NR NR Yes No
CE
Stojovska-
Siemens Magnetom Avanto 1.5 CE A,C,S 6/2 20/40/60 Yes No Yes
Jovanovska E [29]
30/60/90/1
Takeuchi M [30] Philips Gyroscan Intera 1.5 T2/DWI/CE A,S 4/0.4 A,S 4/0.4 0,1000 P 3/0 Yes Yes Yes
20
Tillou X [31] GE NR 1.5 T1/T2/CE Yes Yes Yes
A, O
Wang HJ [5] Siemens Magnetom Trio 3.0 DWI 4/0.4 0,1000 NR Yes Yes
(P)
Watanabe H [32] Philips Intera Achieva 1.5 T1/T2/CE A,S 5/2 P 5/0 30/95/120 NR Some Yes

Wu LM [33] GE Signa HDx 3 T2/DWI A,S 5/1 A 3/1 0,1500 Yes Yes Yes
T1/T2/DWI/
Yamada Y [34] GE Signa Excite HDxt 1.5 A,C,S 5/2 A 5/2 0,1000 A,C,S 5/2 NR Yes NR No
CE
A = axial; C= coronal; CE = contrast-enhanced; DWI = diffusion-weighted imaging; DTI = diffusion-tensor imaging; NR = not reported; OS
= oblique sagittal; P = perpendicular to tumor base; ST = slice thickness; TUR = transurethral resection; T1 = T1-weighted imaging; T2 = T2-
weighted imaging; S = sagittal; STIR = short tau inversion recovery