- Email: [email protected]
Dynamic Contrast-Enhanced Magnetic Resonance Imaging as a Diagnostic Tool in the Assessment of Tumour Angiogenesis in Urinary Bladder Cancer
Canadian Association of Radiologists Journal xx (2019) 1e10 www.carjonline.org
Magnetic Resonance Imaging / Formation image de resonance magnetique
Dynamic Contrast-Enhanced Magnetic Resonance Imaging as a Diagnostic Tool in the Assessment of Tumour Angiogenesis in Urinary Bladder Cancer Omar Ahmad Hassanien, MDa,*, Rasha T. Abouelkheir, Mscb, Mohamed Ibrahim Abou El-Ghar, MDc, Manal Ezzat Badawy, MDa, Samir Abdel-hakim El Gamal, MDa, Mohamed Abd El-Hamid, MScc a
b
Faculty of Medicine, Tanta University, Tanta, Gharbia Governorate, Egypt Diagnostic Radiology Department, Urology and Nephrology Center, Mansoura University, Mansoura, Dakahlia Governorate, Egypt c Urology and Nephrology Center, Faculty of Medicine, Mansoura University, Mansoura, Dakahlia Governorate, Egypt
Abstract Purpose: The aim of study is to assess the role of dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) and correlation with tumour angiogenesis in evaluation of urinary bladder cancer. Material and Methods: The study included 81 patients with recent presumed diagnosis of bladder tumour or who came for follow up after management of histopathologically proven bladder cancer. All had DCE-MRI with timeesignal intensity curve. The radiologic results then correlated with the histopathologic results using both haematoxylin and eosin stain and immuno-histochemical staining for localization and evaluation of CD34 immunoreactivity as a detector for the microvessel density (MVD) and tumour angiogenesis. Results: Seventy-one cases were pathologically proven to be malignant: 41 cases (58%) showed type III timeesignal intensity curve (descending); 22 cases (31%) showed type II (plateau); and 8 cases (11%) showed type I (ascending) curve. The sensitivity of DCE-MRI in stage T1 bladder tumour was 80%; in stage T2, it was (90.9%); and in stage T3, it was (96.9%). Overall accuracy of DCE-MRI in tumour staging was 89.5% and P ¼ .001 (significant). Values more than the cutoff value ¼ 76.13 MVD are cystitis with sensitivity (90%), specificity (91%), and P value is .001, which is statistically highly significant. Conclusion: There is a strong positive association between DCE-MRI (staging and washout slope of the timeesignal intensity curve) with histopathologic grade, tumour stage, and MVD in bladder cancer. So, DCE-MRI can be used as reliable technique in preoperative predictions of tumour behavior and affect the planning of antiangiogenetic therapy. Resume Objectif : L’objectif de cette etude est d’evaluer le r^ole de l’imagerie par resonance magnetique dynamique de contraste (IRM-DSC) et la correlation de cette technique avec l’angiogenese tumorale dans l’evaluation du cancer de la vessie. Materiel et methodes : Cette etude portait sur 81 patients presentant un diagnostic recent presume de tumeur de la vessie ou en consultation de suivi consecutive a la prise en charge d’un cancer de la vessie au diagnostic prouve par histopathologie. Ils ont tous subi une IRM-DSC avec courbe temporelle de la variation d’intensite de signal. Les resultats radiologiques ont ensuite ete confrontes aux resultats histopathologiques obtenus par coloration a l’hematoxyline et a l’eosine combinee au marquage immunohistochimique pour la localisation et l’evaluation de l’immunoreactivite au CD34, en tant qu’outil de detection de la densite microvasculaire (DMV) et de l’angiogenese tumorale. Resultats : Les resultats pathologiques ont mis en evidence 71 cas de malignite: pour 41 cas (58 %), la courbe temporelle de la variation d’intensite de signal etait de type III (descendante); pour 22 cas (31 %), elle etait de type II (plateau); et pour 8 cas (11 %), de type I (ascendante). La sensibilite de l’IRM-DSC pour les tumeurs de la vessie de stade T1 etait de 80,0 %; au stade T2, elle etait de 90,9 %; et au stade T3, de 96,9 %. L’exactitude globale de l’IRM-DSC pour la detection du stade tumoral etait de 89,5 %; P ¼ 0,001 (significatif). Les valeurs de densite microvasculaire superieures a la valeur seuil etaient egales a 76,13, ce qui correspond a une cystite, avec une sensibilite de 90 %, une specificite de 91 % et une valeur P de 0,001, ce qui est statistiquement hautement significatif.
* Address for correspondence: Dr. Omar A. Hassanien, Faculty of Medicine, Tanta University, Tanta, Gharbia Governorate, Egypt 31527.
E-mail address: [email protected] (O. A. Hassanien).
0846-5371/$ - see front matter Ó 2018 Canadian Association of Radiologists. All rights reserved. https://doi.org/10.1016/j.carj.2018.11.004
2
O. A. Hassanien et al. / Canadian Association of Radiologists Journal xx (2019) 1e10
Conclusion : Une forte association positive entre l’IRM-DSC (type de rehaussement [stadification] et pente de decroissance [elimination ou washout] de la courbe temporelle d’intensite de signal) avec le grade histopathologique, le stade tumoral et la valeur de DMV du cancer de la vessie a ete mise en evidence. Ainsi, l’IRM-DSC peut constituer une technique fiable utilisee dans les predictions preoperatoires du comportement tumoral et avoir un impact sur la planification de la therapie anti-angiogenetique. Ó 2018 Canadian Association of Radiologists. All rights reserved. Key Words: Urinary bladder cancer; Tumoral angiogenesis; Dynamic contrast magnetic resonance imaging
Urinary bladder cancer is considered the most common malignancy of the urinary system. It is the eighth leading cause of cancer-related death in the United States [1]. Many factors were proved to play an important role as predisposing and risk factors in the pathogenesis of the urinary bladder cancers. Some factors are related to occupation (eg, exposure to organophosphate-based pesticides), others due to genetic predisposition or habits (cigarette smoking). Schistosomiasis, pelvic irradiation, chronic cystitis, and human papilloma virus are also other risk factors and vary between different populations [2e4]. Recent research stated that angiogenesis is the main step in the pathophysiology of malignant tumours and their potential for growth and spreading [5]. The rate of growth of bladder tumours is parallel to the presence of a rich blood supply formed by a large network of immature disturbed blood vessels in addition to the density of this network [6]. The intra-tumoural microvessel density (MVD), which is a measure of tumour angiogenesis in bladder cancer, was proved to correlate with tumoural malignancy [7], and its increase is associated with poorer a survival rate and increased rates of recurrence and mitotic index [8]. On the other hand, magnetic resonance imaging (MRI) has been shown to be superior to other imaging modalities in the diagnosis of bladder neoplasms, especially of superficial and multiple tumours and in tumour staging due to its high soft tissue discrimination [9]. Dynamic contrast-enhanced MRI (DCE-MRI) using fast MRI sequences obtained before, during, and after the rapid intravenous administration of gadolinium gives valuable data regarding both morphologic changes and physiological alteration associated with bladder malignancy [10]. The patterns of contrast enhancement on DCE-MRI are likely related to tumour angiogenesis. It could be used to assess the therapeutic response in patients who have not undergone surgical removal of their tumours [9]. We aimed at this study to assess the role of DCE-MRI and its correlation with tumour angiogenesis in the evaluation of urinary bladder cancer. Patients and Methods After approval of the Institutional Review Board of the Faculty of Medicine of Tanta University in June 2013, this study was performed at the Urology and Nephrology Center at Mansoura University between November 2013-June 2016, including a 1-year follow-up period. Ninety-four patients were initially included in the study, from whom 13 patients were excluded, t3 patients did not complete the study due to
claustrophobia, 4 patients were excluded due to bad image quality and motion artifacts, and 6 patients did not undergo biopsy, so we had no pathologic correlation for them. Finally, 81 patients were included in the study. Ethics committee approval, informed written consent, and explanation of the MRI procedure to the patients were completed. Privacy and confidentiality of all patient data were guaranteed, and there was a code number for every patient’s file that includes all investigations. All data provision were monitored and used for scientific purposes only. The cases were selected among the following: A. Patients with recent presumed diagnosis of bladder tumour at the outpatient clinic. In our country we have a high incidence of bladder cancer. We depend on the patient’s history of hematuria and lower urinary tract symptoms in suspect cases of bladder lesion. Then, we do outpatient cystoscopy for patients screening. B. Patients who came for follow-up after the management of histopathologically proven bladder cancer. All patients were subjected to the following: 1. Informed consent after full explanation of the benefits and risks of the procedure. 2. Full history taking: pelvic pain, painless gross hematuria, smoking, history of transurethral resection of bladder tumor (TURBT), and irritative bladder symptoms as dysuria, urgency, or frequency of urination. 3. Clinical examination: general signs of malignancy and metastasis, local examination by digital rectal or pervaginal examination to assess the mass, and its mobility or fixation to the pelvic wall. 4. Laboratory investigation: serum creatinine and urine cytology. 5. DCE-MRI within 3 weeks of initial diagnosis. The radiologic results then correlated with the histopathologic results. 6. Conventional cystoscopy. 7. Histopathologic investigation. Exclusion criteria included patients with renal failure, impaired renal function with glomerular filtration rate < 30 mL/min/1.73 m2 due to the risk of nephrogenic systemic fibrosis; past history of allergy from the MRI contrast media; artificial metallic prosthesis, eg, ‘‘artificial valves or cardiac pacemakers,’’ patients who refuse the study and those with severe claustrophobia.
DCE-MRI as diagnostic tool / Canadian Association of Radiologists Journal xx (2019) 1e10
MRI Protocol Patients were instructed to begin drinking water 1 hour before the MRI study and to present with a full bladder. In 8 patients with a urethral catheter, 250e400 mL sterile saline was used to distend the bladder. During the imaging procedure, fullness of the bladder was checked at localizer images and the examination was delayed if the bladder was not full. All MRI studies were done on a 1.5 Tesla Signa Horizon LX Echospeed scanner (GE Medical Systems, Milwaukee, WI) using a phased array pelvic coil. The procedure began by obtaining a localizer (scout image) to identify the pelvis. The following sequences were performed: A. T2-weighted high resolution sequences of the pelvis in axial, sagittal, and coronal directions to locate the lesions and to reveal their morphologic characteristics using the following parameters: time to repetition ¼ 7000e8000 ms, time to echo ¼ 90e 102 ms, band width ¼ 20e83 kHz, 256 256 matrix, slice thickness of 3 mm, intersection gap of 0.5 mm, and a field of view ¼ 20 20 cm. B. DCE-MRI: It was performed using fast spoiled gradient T1 sequence (pre- and post-contrast) in the axial, coronal, and sagittal planes by using the following parameters: repetition time ¼ 68 ms, time to echo ¼ 2 ms, 5 mm thickness, no inter-slice gap, flip angle 60 degrees, field of view 32 32 cm, and matrix 256 160. Images obtained before and immediately after manually injected gadolinium (gadoteric acid [Dotarem 0.5 mmoL/ mL]) at a dose of 0.1 mmol/kg of body weight (maximum, 20 mL) intravenously through the antecubital vein, followed immediately by a flush of 20 mL of saline solution. We tried to be fast as possible through a wide bore cannula to keep time less than 10 seconds for contrast and the same time for saline. Twelve sequential gradient T1 series, each lasting 15 seconds, were performed in 3 minutes.
3
hypointensity and enhances late. We used the following staging system according to Takeuchi et al [11]. Stage T1 Intact muscle layer at the base of the tumour that shows low signal intensity on T2-weighted MRI and no early enhancement on DCE-MRI. Stage T2 Disrupted bladder wall muscle’s hypointense line and early enhancement of the lesion without perivesical fat infiltration. Stage T3 Lesion with an irregular shaggy outer border and streaky areas in perivesical fat of the same signal intensity as the tumour. Stage T4 Lesion extending into an adjacent organ or abdominal and pelvic side walls. TimeeSI Curves TimeeSI curves were also constructed from signal intensity values obtained from freely drawn regions of interest selected on the basis of optimal visualization of the lesion and the region of greatest enhancement [12]. Type I (Ascending) Curve Time-SI curve shows enhancement, followed by a slow increase. Type II (Plateau) Curve
Image Post Processing Time-SI curve shows enhancement, followed by a plateau. The data were processed by using the software installed on the workstation system provided by the machine vendor (Advantage Workstation, version 4.6 [GE Healthcare, Bue, France]). The region of interest (ROI) for MRI enhancement was obtained from the region of maximum enhancement in the tumour that correlate with the areas of high vascularity with generation of time-signal intensity (SI) curves. The sizes of the ROIs were between 1-2 cm2. Three ROIs were placed within the masses for SI calculation, and we chose the highest value.
Type III (Descending) Curve Time-SI curve shows enhancement, followed by washout. Final diagnosis was confirmed on histopathology of the operated specimen. Staging and grading of tumours were analyzed on MRI and compared with the operative and histopathologic findings. Analysis of time-SI curves was done in all cases. Histopathologic Analysis
Image Analysis With DCE-MRI, bladder tumours, mucosa, and submucosa enhance early, but the muscle layer maintains its
Routinely processed, paraffin-embedded bladder tissues were used in this study. Three to 4 microns thick serial sections mounted on coated slides with Histogrip solution
4
O. A. Hassanien et al. / Canadian Association of Radiologists Journal xx (2019) 1e10
(Zymed Laboratories, Inc, San Francisco, CA), and subjected to: (1) haematoxylin and eosin stain for routine histopathologic examination, typing, grading, and staging of tumour; (2) immunohistochemical ‘‘IHC’’ staining for localization and evaluation of CD34 immunoreactivity. The CD34 is an endothelial cell marker, Ab-1 (Clone QBEnd/10). Mouse monoclonal antibody. (NeoMarkers, Inc) to detect the MVD. The MVD, defined as capillaries and small venules, was assessed in areas with solid tumour morphology away from any artifact or necrosis and without prior knowledge of patient outcome. After immuno-histochemical staining with CD34, the vessel counting was performed in 4 areas of maximal neovascularization where the highest number of discrete microvessels was stained (hot spots). Low power light microscopy at 40 magnification was used to scan the often heterogeneous tumour sections for these areas. At 250 magnification, counts were made of all distinct brown staining endothelial cells or cell clumps. The MVD was defined as the mean of the highest 4 counts. On the basis of the MRI findings, specimen sectioning for pathologic examination was performed without knowledge of patients’ outcome by neither the pathologist nor the radiologist. Statistical Data Data were analyzed using SPSS version 21.0 (IBM). Diagnostic efficacy was depicted in terms of sensitivity, specificity, and overall accuracy. The level of confidence was kept at 95%, hence a P < .05 indicated a significant association. Sensitivity, specificity, and accuracy of DCE-MRI were assessed on a stage-by-stage basis, and the gold standard was the histopathologic confirmation in all cases. We used the following statistical methods: A. Distribution of data was analyzed by mean and standard deviation. B. Frequency tables were analyzed using c2 test. C. Differences in between the groups were analyzed using independent-samples tests. D. Reported P values were obtained by using two-sided tests, and in all analysis, P < .05 was regarded as the limit of statistical significance. E. Receiver operating characteristic (ROC) curves were obtained for measuring the sensitivity and specificity of different variables. F. Kappa statistics were calculated to determine the strength of agreement and the harmony between the DCE-MRI stage and the histopathologic. stage. Value of kappa (< 0.20: poor), (0.21e0.40: fair), (0.41e0.60: moderate), (0.61e0.80: good), and (0.81e1.00: almost perfect). G. Spearman’s correlation test was used to determine the strength of the relationships between DCE-MRI results, MVD, and histopathologic and stage and grade (r ¼ 0.0e0.2 ¼ No or very low correlation),
Table 1 Number and classification of the 81 cases according to their main complaint Number of Cases Complaint Hematuria Follow-up post TURBT Irritative bladder symptoms Incidental Total
Benign 3 (30%) 3 (30%) 4 (40%) 10 (100%)
Malignant 58 9 2 2 71
(81.7%) (12.7%) (2.8%) (2.8%) (100%)
TURBT ¼ transurethral resection of bladder tumor.
(r ¼ 0.2e0.4 ¼ low correlation), (r ¼ 0.4e 0.6 ¼ reasonable correlation), (r ¼ 0.6e0.8 ¼ high correlation), and (r ¼ 0.8e1.0 ¼ very high correlation).
Results The current study included 81 patients (70 male [86.4%] and 11 female [13.6%]). The age ranged between 20-80 years, with mean age 59.25 years 10.167 standard deviation (SD). Seventy-one cases (87.7%) were pathologically proven to be malignant, while the other 10 patients (12.3%) were proved to be benign. Hematuria was the main complain, as it was present in 61 patients (75.3%), while 12 patients (14.8%) came for regular follow-up post TURBT, and 6 patients (7.4%) had irritative bladder symptoms (Table 1). Regarding the 71 malignant tumours, they were pathologically classified according to T staging: T1: 15 cases (21.1%), T2: 22 cases (31%), T3: 32 cases (45.1%), T4: 2 cases (2.8%); one of them was T4a, but was inoperable (the patient was not fit), and the other was T4b. Both were managed by taking chemotherapy, and there was no pathologic staging for them, so staging depended on MRI only in these 2 cases.
MR T2 W SI of the Bladder Lesions Ten cases were pathologically proven to be benign in nature: 6 cases (60%) had intermediate SI, 3 cases (30%) had hypointense SI, and 1 case (10%) had hyperintense SI at T2 W. Seventy-one cases were pathologically proven to be malignant in nature. The most common SI was intermediate: 49 cases (69%), 19 cases (26.8%), had hypointense SI, and 3 cases (4.2%) had hyperintense SI at T2 W (Table 2). Table 2 Classification of the cases according to MR T2 W SI of the bladder lesions T2 W SI
Benign
Malignant
Intermediate Hypointense Hyperintense
6 (60%) 3 (30%) 1 (10%)
49 (69%) 19 (26.8%) 3 (4.2%)
MR ¼ magnetic resonance; SI ¼ signal intensity.
DCE-MRI as diagnostic tool / Canadian Association of Radiologists Journal xx (2019) 1e10
5
Table 3 Type of SI curve in correlating with malignant pathologic staging Type of the curve Pathologic staging
Type I (Ascending)
Type II (Plateau)
Type III (Descending)
Total
T1 T2 T3 T4 (MR staging) Total
4 2 1 1 8
11 10 1 22
10 30 1 41
15 22 32 2 71
(26.7%) (9.1%) (3.1%) (50%) (11%)
(73.3%) (45.5%) (3.1%) (31%)
(45.5%) (93.8%) (50%) (58%)
(100%) (100%) (100%) (100%) (100%)
MR ¼ magnetic resonance; SI ¼ signal intensity.
Figure 1. 31-year-old female patient, presented with irritative bladder symptoms and frequent attacks of hematuria 1 year ago, she had no history of previous transurethral resection of bladder tumor. Proved to be follicular cystitis. Magnetic resonance imaging T2 high resolution: axial (A), coronal (B), and sagittal (C) views revealed: reduced bladder capacity, irregular circumferential bladder wall thickening with sessile lesions at the bladder dome, anterior, and right lateral walls, displaying hypointense signal intensity with infiltration of the perivesical fat (red arrows). Dynamic contrast-enhanced magnetic resonance imaging, axial view. (D) Early post-contrast gradient T1 showed mild diffuse enhancing circumferential bladder thickening (yellow arrow). (E) Late post-contrast gradient T1: persistent and increasing enhancement (yellow arrow). (F) Timeesignal intensity curve shows slow enhancement followed by slow increase with no washout. (G) Haematoxylin-eosin stained slide: follicular cystitis. (H) CD34 expression (250). Mean of MVD ¼ 84.
6
O. A. Hassanien et al. / Canadian Association of Radiologists Journal xx (2019) 1e10
Figure 2. 57-year-old male patient with cancer bladder, presented with hematuria 4 months ago, he had no previous history of transurethral resection of bladder tumor. Magnetic resonance imaging T2 high resolution: axial (A), coronal (B), and sagittal (C) views revealed: intra-vesical fungating soft tissue mass, displays intermediate signal intensity (SI), arising from the left lateral bladder wall (red arrow), encroaching upon the left ureteric orifice with intact hypointense SI of the related muscle layer. Dynamic contrast-enhanced magnetic resonance imaging revealed, axial view. (D): Early post-contrast gradient T1 showed an intra-vesical soft tissue mass arising from the left lateral bladder wall, encroaching upon the left ureteric orifice. The mass showed earlier enhancement than the muscle layer (yellow arrow). (E) Late post-contrast gradient T1 sequence: The detrusor muscle is seen intact and showed delayed enhancement (yellow arrow). Persistent tumour enhancement is also noted. No extra-vesical fat infiltration. (F) TimeeSI curve shows rapid enhancement followed by plateau. Stage T1. (G) Haematoxylin-eosin stained slide showed grade II papillary TCC. Stage T1. (H) CD34 expression (250). Mean of MVD ¼ 37.
Types of TimeeSI Curve for the 81 Patients Included in the Study DCE-MRI results according to the washout slope of the time-SI curve revealed: 17 cases (21%) showed ascending curve (type I), 23 cases (28%) showed plateau curve (type II), and 41 cases (51%) showed descending curve (type III). In the 71 cases pathologically proven to be malignant, the type of the time-SI curve was as follows: 41 cases (58%) showed type III (descending) curve, 22 cases (31%) showed type II (plateau), and 8 cases (11%) showed type I
(ascending) curve (Table 3). Ten cases proved to be cystitis: 9 cases (90%) showed type I (ascending) curve and 1 case (10%) showed type II (plateau) curve (Figure 1).
Type of Curve in Correlation With Pathologic Tumour Staging Fifteen cases proved to be stage T1: 4 cases (26.7%) showed type I (ascending) curve and 11 cases (73.3%) showed type II (plateau) curve (Figure 2).
DCE-MRI as diagnostic tool / Canadian Association of Radiologists Journal xx (2019) 1e10
7
Figure 3. 52-year-old male patient with cancer bladder, presented with hematuria 6 months ago; he had no previous history of transurethral resection of bladder tumor. Magnetic resonance imaging T2 high resolution: axial (A), coronal (B), and sagittal (C) views revealed: intra-vesical fungating soft tissue mass, displays intermediate signal intensity (SI), arising from the left postero-lateral bladder wall (red arrows), encroaching upon the bladder neck and left ureteric orifice with loss of hypointense SI of the related muscle layer. Dynamic contrast-enhanced magnetic resonance imaging revealed: axial (D) early post-contrast gradient T1 showed earlier enhancement of the mass than the muscle layer (yellow arrow). (E) Late post-contrast gradient T1 sequence: the detrusor muscle showed delayed enhancement as well as interruption by the tumour (yellow arrow). No extra-vesical fat infiltration. (F) TimeeSI curve showed rapid enhancement followed by washout. Stage T2. (G) haematoxylin-eosin stained slide showed grade III papillary TCC. Stage T2. (H) CD34 expression (250): mean of MVD ¼ 68.
Twenty-two cases proved to be stage T2: 2 cases (9.1%) showed type I (ascending) curve, 10 cases (45.5%) type II (plateau) curve, and 10 cases showed (45.5%) type III (descending) curve (Figure 3). Thirty-two cases proved to be stage T3: 1 case (3.1%) showed type I (ascending) curve, 1 case (3.1%) showed type II (plateau) curve, and 30 cases (93.8%) showed type III (descending) curve. Two cases proved to be stage T4: 1 case (50%) showed type I (ascending) curve, and 1 case (50%) showed type III (descending) curve (Table 3). The sensitivity of dynamic MRI in stage T1 bladder tumour was 80%; in stage T2, it was (90.9%); and in stage T3, it was (96.9%). In T4 bladder tumours, one was stage T4a
(inoperable, as the patient was unfit for operation), and the other was T4b, so management depended upon MRI staging, by taking chemotherapy for both cases. The overall accuracy of DCE-MRI in tumour staging was 89.5%. Over-staging occurred in 6 cases (7.4%): 3 cases T1, 2 cases T2, and 1 case (cystitis). Under-staging occurred in 1 case, which pathologically diagnosed as T3, and we misdiagnosed it as cystitis (Table 4). Kappa Statistics To determine the level of agreement and the harmony between the DCE-MRI stage and the histo-pathologic stage,
8
O. A. Hassanien et al. / Canadian Association of Radiologists Journal xx (2019) 1e10
Table 4 Sensitivity values of dynamic MRI compared with pathologic staging Number of cases detected by dynamic MRI/total number detected by pathology
Pathologic Staging
Sensitivity
T1 12/15 80%
T2 20/22 90.9%
T3 31/32 96.9%
Cystitis 9/10 90%
Overall accuracy
P value
k statistics
Spearman correlation test
89.5%
P ¼ .001
k ¼ 0.880
r ¼ 0.965
P < .05 was regarded as the limit of statistical significance. Kappa statistics: k ¼ 0.81e1.00 is almost perfect. Spearman’s correlation test: r ¼ 0.8e1.0 ¼ very high correlation. MRI ¼ magnetic resonance imaging.
DCE-MRI agreement was almost perfect in assigning the tumour stage (k ¼ 0.880, P ¼ .001).
staging (r ¼ 0.606, P ¼ .001) where (r ¼ 0.6e0.8 ¼ high correlation).
Correlation Between the DCE-MRI Staging and Histopathologic
Correlation Between Dynamic MRI Curves and MVDs
Staging (Spearman correlation test) A very high positive correlation was noted between the DCE-MRI staging and histopathologic staging (r ¼ 0.965) where r ¼ 0.8e1.0 ¼ very high correlation (Table 4). Correlation Between the Type of SI Curve and Histopathologic Staging High positive correlation between the type of curve and pathologic staging (r ¼ 0.646, P ¼ .001) where r ¼ 0.6e0.8 ¼ high correlation. Tumour Angiogenesis
Spearman Correlation test showed reasonable positive correlation between the washout slope of dynamic MRI curve and mean MVD (r ¼ 0.471, P ¼ .001), where r ¼ 0.4e0.6 is reasonable correlation and P ¼ .001 is statistically highly significant. Discussion The formation of new blood vessels from pre-existing vasculature (angiogenesis) is a critical part of tumour growth and progression [13,14]. Tumour growth is affected by inhibition of angiogenesis, and this may even result in tumour regression [15]. The majority of published studies have shown a positive correlation between intra-tumoural
Microvessel densities were found to be between 17-122 vessels/area (mean ¼ 55 23.3 SD vessels/area). The mean of MVD in the 10 benign cases was 77.63 16.73 SD. The mean of MVD of the 71 malignant cases was 53.77 22.36 SD. Values more than the cutoff value ¼ 76.13 MVD are cystitis. The area under curve is 0.863 (95% confidence interval 0.723e1.000). Sensitivity is 90%. Specificity is 91%. P value is .001, which is statistically highly significant (Figure 4). MVD Correlating With Pathologic Tumour Staging Regarding the tumour staging, in the 15 cases were diagnosed as T1, the mean of the MVD was 27.12 7.14 SD. In the 22 cases diagnosed as T2, the mean of MVD was 50.13 16.78 SD. In 32 cases diagnosed as T3, the mean of MVD was 65.72 19.65 SD. In 2 cases (staged by MRI) as T4, the mean of MVD was in case diagnosed as T4a was 72, and in the other case diagnosed as T4b, mean of MVD was 21 (Table 5). Spearman correlation test was done to determine the strength of the relationship between the mean MVD and tumour staging: There is a high positive correlation was also noted between the mean MVD and tumour pathologic
Figure 4. Receiver operating characteristic (ROC) curve: sensitivity and specificity of microvessel density in differentiation between type of cells (benign and malignant cells).
DCE-MRI as diagnostic tool / Canadian Association of Radiologists Journal xx (2019) 1e10 Table 5 Mean microvessel density correlating with pathologic tumour staging Pathologic staging
Number
Mean of MVD
SD
T1 T2 T3
15 22 32
27.12 50.13 65.72
7.14 16.78 19.65
MVD ¼ microvessel density; SD ¼ standard deviation.
MVD, ‘‘a measure of tumour angiogenesis’’ and prognosis in bladder tumours [7,16e18]. Radiological evaluation is a significant part of diagnosis and staging of bladder cancer. Functional dynamic contrast-enhanced MRI can assess the microcirculation and visualize the neo-angiogenesis of malignant tissues via the dynamic signal enhancement of a contrast agent [19,20]. This study was carried out with 81 patients (70 men and 11 women); 71 cases were proven to be malignant. Mean age was 60 years; this copes with the recorded age and sex incidence of bladder tumours [21,22]. In the current study, the sensitivity of dynamic MRI in in the radiological staging compared to histopathologic staging in stage T1 bladder tumour was 80%. In stage T2, it was 90.9%, and in stage T3, was (96.9%), with overall accuracy 89.5%. Two cases with T4 bladder tumours were present in this study; one was stage T4a (inoperable, as the patient was unfit for operation), and the other was T4b, so management depended upon MRI staging, by taking chemotherapy for both cases. Our study showed higher sensitivity than that of Gupta et al [23]. In their study, the sensitivities of dynamic MRI in stage (T1, T2, T3, T4) bladder tumour were 62.5%, 83.3%, 50%, and 100%, respectively. Over-staging was seen in 6 patients (7.4%) in this study, compared to study by Gupta et al [23] where over-staging occurred in 20% of cases. The reason of overestimation could be due to abnormal signals in perivesical fat, especially if associated with cystitis that could be incorrectly identified as tumour invasion. Vascular hyperemia adjacent to the tumour could cause these abnormal signals. On the other hand, under-staging occurred in 1 case, which pathologically diagnosed as T3, and we misdiagnosed it as cystitis. This case has previous transurethral resection before reference to our center, and the mass could not be visualized among the thickened bladder wall. In our study, statistically, the extent of agreement between dynamic contrast MRI, and histopathologic staging was almost perfect (k ¼ 0.880) and highly significant (P ¼ .001) with positive very high correlation (r ¼ 0.965). These results are higher compared to the study done by both of Gupta et al [23] on 60 patients that resulted in good agreement equals (k ¼ 0.690) and significant (P < .001), and with Tuncbilek et al [9] on 24 patients that resulted in fair agreement (k ¼ 0.365) and significant (P ¼ .002). It must be emphasized that MVD evaluation under the microscope is somewhat subjective and inter-observer variation can be substantia [7]. So, we proceeded for the data
9
analysis of the angiogenesis of the bladder lesions and correlating them with the type of cells, tumour stage, grade, and the incidence of recurrence. We studied the correlation between the MVD and type of cells ‘‘benign and malignant’’; we noticed the higher vascularity of cases of cystitis, compared to the vascularity of malignant tumours. This could be explained by the higher vascularity of the active cells, edema, inflammatory infiltrate, and vascular congestion during the inflammatory process. To our knowledge, our study was the first to correlate the mean of MVD with the type of cells (cystitis vs malignant). From the ROC curve, by applying the cutoff value ¼ 76.13 MVD, the higher MVD above the cutoff value are cystitis with sensitivity (90%), specificity (91%). P value (.001), which is statistically highly significant. We studied the correlation between the MVD and pathologic tumour staging that revealed high positive correlation between MVD and tumour stage (r ¼ 0.606, P ¼ .001). In the current study, angiogenesis increases in parallel with increasing tumour stage and grade, apart from T4 (2 cases), due to small group number; also, one of them was diagnosed as undifferentiated cell carcinoma with large areas of cystic necrosis, so it could explain the lower MVD than the T2 and T3 tumours. This is fully consonant with the observations by Gehani et al [7] and Canoglu et al [24] that also confirmed that MVD in bladder carcinoma correlates with the tumour grade, stage, and its malignant potential. Our study revealed that there is a reasonable positive correlation was also noted between the washout slope of dynamic MRI curve and mean MVD (r ¼ 0.471, P ¼ .001) that copes with the study done by Tuncbilek et al [9]. The limitations of our study were related to the technique of dynamic MRI using the manual technique of injection, which may affect the results; also, the short follow-up period of patients. At the end of this study, we can conclude that DCE-MRI supplemented with time-SI curve is a promising non-invasive diagnostic tool in the staging of cancer bladder through the tumoural vascular behavior and angiogensis. This may allow DCE-MRI to be used in preoperative predictions of tumour behavior and biologic activity that can affect the therapy planning with antiangiogenetic therapy. We recommend a larger and longer study to confirm the findings. References [1] Gang J, Fan J, Wang Q, et al. Decreased REG1 a expression suppresses growth, invasion and angiogenesis of bladder cancer. Eur J Surg Oncol 2017;43:837e46. [2] Zarzour AH, Selim M, Abd-Elsayed AA, et al. Muscle invasive bladder cancer in Upper Egypt: the shift in risk factors and tumor characteristics. BMC Cancer 2008;8:250. [3] Felix AS, Soliman AS, Khaled H, et al. The changing patterns of bladder cancer in Egypt over the past 26 years. Cancer Causes Control 2008;19: 421e9. [4] Salem HK, Mahfouz S. Changing patterns (age, incidence, and pathologic types) of schistosoma-associated bladder cancer in Egypt in the past decade. Urology 2012;79:379e83.
10
O. A. Hassanien et al. / Canadian Association of Radiologists Journal xx (2019) 1e10
[5] T€ urkbey B, Thomasson D, Pang Y, et al. The role of dynamic contrastenhanced MRI in cancer diagnosis and treatment. Diagn Interv Radiol 2010;16:186e92. [6] Xie Q, Zhang J, Wu P-H, et al. Bladder transitional cell carcinoma: correlation of contrast enhancement on computed tomography with histological grade and tumour angiogenesis. Clin Radiol 2005;60:215e23. [7] Gehani K, Al-Kikhia L, Mansuri N, et al. Angiogenesis in urinary bladder carcinoma as defined by microvessel density (MVD) after immunohistochemical staining for Factor VIII and CD31. Libyan J Med 2011;6. [8] Hall A. The role of angiogenesis in cancer. Compar Clin Pathol 2005; 13:95e9. [9] Tuncbilek N, Kaplan M, Altaner S, et al. Value of dynamic contrastenhanced MRI and correlation with tumor angiogenesis in bladder cancer. AJR Am J Roentgenol 2009;192:949e55. [10] Choyke PL, Dwyer AJ, Knopp MV. Functional tumor imaging with dynamic contrast-enhanced magnetic resonance imaging. J Magn Reson Imaging 2003;17:509e20. [11] Takeuchi M, Sasaki S, Ito M, et al. Urinary bladder cancer: diffusionweighted MR imaging-accuracy for diagnosing T stage and estimating histologic grade. Radiology 2009;251:112e21. [12] Cheng L, Li X. Breast magnetic resonance imaging: kinetic curve assessment. Gland Surg 2013;2:50e3. [13] Iagaru A, Gambhir SS. Imaging tumor angiogenesis: the road to clinical utility. AJR Am J Roentgenol 2013;201:183e91. [14] Folkman J. Role of angiogenesis in tumour growth and metastasis. Semin Oncol 2002;29:15e8. [15] Turkbey B, Kobayashi H, Ogawa M, et al. Imaging of tumor angiogenesis: functional or targeted? AJR Am J Roentgenol 2009;193: 304e13.
[16] Chikazawa M, Inoue K, Fukata S, et al. Expression of angiogenesisrelated genes regulates different steps in the process of tumor growth and metastasis in human urothelial cell carcinoma of the urinary bladder. Pathobiology 2008;75:335e45. [17] Sarı A, Ermete M, Dag F, et al. Evaluation of angiogenesis in urothelial carcinoma of bladder and its relation with prognostic factors. Ege Journal of Medicine 2007;46:135e40. [18] Goddard JC, Sutton CD, Furness PN, et al. Microvessel density at presentation predicts subsequent muscle invasion in superficial bladder cancer. Clin Cancer Res 2003;9:2583e6. [19] Tekes A, Kamel I, Imam K, et al. Dynamic MRI of bladder cancer: evaluation of staging accuracy. AJR Am J Roentgenol 2005;184: 121e7. [20] Rose T, Lotan Y. Advancements in optical techniques and imaging in the diagnosis and management of bladder cancer. Urol Oncol 2018;36: 97e102. [21] Epstein JI, Lotan TL. The lower urinary tract and male genital system. In: Kumar V, Abbas A, Aster JC, editors. Robbins & Cotran Pathologic Basis of Disease. 9th ed. Toronto: Saunders Elsevier; 2015. p. 959e90. [22] Rabie E, Faeghi F, Izadpanahi MH, et al. Role of dynamic contrastenhanced magnetic resonance imaging in staging of bladder cancer. J Clin Diagn Res 2016;10:TC01e5. [23] Gupta N, Sureka B, Kumar MM, et al. Comparison of dynamic contrast-enhanced and diffusion weighted magnetic resonance image in staging and grading of carcinoma bladder with histopathological correlation. Urol Ann 2015;7:199e204. [24] Canoglu A, G€og€us¸ C uk Y, et al. Microvessel density as a prog¸ , Bed€ nostic marker in bladder carcinoma: correlation with tumor grade, stage and prognosis. Int Urol Nephrol 2004;36:401e5.
Magnetic Resonance Imaging / Formation image de resonance magnetique
Dynamic Contrast-Enhanced Magnetic Resonance Imaging as a Diagnostic Tool in the Assessment of Tumour Angiogenesis in Urinary Bladder Cancer Omar Ahmad Hassanien, MDa,*, Rasha T. Abouelkheir, Mscb, Mohamed Ibrahim Abou El-Ghar, MDc, Manal Ezzat Badawy, MDa, Samir Abdel-hakim El Gamal, MDa, Mohamed Abd El-Hamid, MScc a
b
Faculty of Medicine, Tanta University, Tanta, Gharbia Governorate, Egypt Diagnostic Radiology Department, Urology and Nephrology Center, Mansoura University, Mansoura, Dakahlia Governorate, Egypt c Urology and Nephrology Center, Faculty of Medicine, Mansoura University, Mansoura, Dakahlia Governorate, Egypt
Abstract Purpose: The aim of study is to assess the role of dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) and correlation with tumour angiogenesis in evaluation of urinary bladder cancer. Material and Methods: The study included 81 patients with recent presumed diagnosis of bladder tumour or who came for follow up after management of histopathologically proven bladder cancer. All had DCE-MRI with timeesignal intensity curve. The radiologic results then correlated with the histopathologic results using both haematoxylin and eosin stain and immuno-histochemical staining for localization and evaluation of CD34 immunoreactivity as a detector for the microvessel density (MVD) and tumour angiogenesis. Results: Seventy-one cases were pathologically proven to be malignant: 41 cases (58%) showed type III timeesignal intensity curve (descending); 22 cases (31%) showed type II (plateau); and 8 cases (11%) showed type I (ascending) curve. The sensitivity of DCE-MRI in stage T1 bladder tumour was 80%; in stage T2, it was (90.9%); and in stage T3, it was (96.9%). Overall accuracy of DCE-MRI in tumour staging was 89.5% and P ¼ .001 (significant). Values more than the cutoff value ¼ 76.13 MVD are cystitis with sensitivity (90%), specificity (91%), and P value is .001, which is statistically highly significant. Conclusion: There is a strong positive association between DCE-MRI (staging and washout slope of the timeesignal intensity curve) with histopathologic grade, tumour stage, and MVD in bladder cancer. So, DCE-MRI can be used as reliable technique in preoperative predictions of tumour behavior and affect the planning of antiangiogenetic therapy. Resume Objectif : L’objectif de cette etude est d’evaluer le r^ole de l’imagerie par resonance magnetique dynamique de contraste (IRM-DSC) et la correlation de cette technique avec l’angiogenese tumorale dans l’evaluation du cancer de la vessie. Materiel et methodes : Cette etude portait sur 81 patients presentant un diagnostic recent presume de tumeur de la vessie ou en consultation de suivi consecutive a la prise en charge d’un cancer de la vessie au diagnostic prouve par histopathologie. Ils ont tous subi une IRM-DSC avec courbe temporelle de la variation d’intensite de signal. Les resultats radiologiques ont ensuite ete confrontes aux resultats histopathologiques obtenus par coloration a l’hematoxyline et a l’eosine combinee au marquage immunohistochimique pour la localisation et l’evaluation de l’immunoreactivite au CD34, en tant qu’outil de detection de la densite microvasculaire (DMV) et de l’angiogenese tumorale. Resultats : Les resultats pathologiques ont mis en evidence 71 cas de malignite: pour 41 cas (58 %), la courbe temporelle de la variation d’intensite de signal etait de type III (descendante); pour 22 cas (31 %), elle etait de type II (plateau); et pour 8 cas (11 %), de type I (ascendante). La sensibilite de l’IRM-DSC pour les tumeurs de la vessie de stade T1 etait de 80,0 %; au stade T2, elle etait de 90,9 %; et au stade T3, de 96,9 %. L’exactitude globale de l’IRM-DSC pour la detection du stade tumoral etait de 89,5 %; P ¼ 0,001 (significatif). Les valeurs de densite microvasculaire superieures a la valeur seuil etaient egales a 76,13, ce qui correspond a une cystite, avec une sensibilite de 90 %, une specificite de 91 % et une valeur P de 0,001, ce qui est statistiquement hautement significatif.
* Address for correspondence: Dr. Omar A. Hassanien, Faculty of Medicine, Tanta University, Tanta, Gharbia Governorate, Egypt 31527.
E-mail address: [email protected] (O. A. Hassanien).
0846-5371/$ - see front matter Ó 2018 Canadian Association of Radiologists. All rights reserved. https://doi.org/10.1016/j.carj.2018.11.004
2
O. A. Hassanien et al. / Canadian Association of Radiologists Journal xx (2019) 1e10
Conclusion : Une forte association positive entre l’IRM-DSC (type de rehaussement [stadification] et pente de decroissance [elimination ou washout] de la courbe temporelle d’intensite de signal) avec le grade histopathologique, le stade tumoral et la valeur de DMV du cancer de la vessie a ete mise en evidence. Ainsi, l’IRM-DSC peut constituer une technique fiable utilisee dans les predictions preoperatoires du comportement tumoral et avoir un impact sur la planification de la therapie anti-angiogenetique. Ó 2018 Canadian Association of Radiologists. All rights reserved. Key Words: Urinary bladder cancer; Tumoral angiogenesis; Dynamic contrast magnetic resonance imaging
Urinary bladder cancer is considered the most common malignancy of the urinary system. It is the eighth leading cause of cancer-related death in the United States [1]. Many factors were proved to play an important role as predisposing and risk factors in the pathogenesis of the urinary bladder cancers. Some factors are related to occupation (eg, exposure to organophosphate-based pesticides), others due to genetic predisposition or habits (cigarette smoking). Schistosomiasis, pelvic irradiation, chronic cystitis, and human papilloma virus are also other risk factors and vary between different populations [2e4]. Recent research stated that angiogenesis is the main step in the pathophysiology of malignant tumours and their potential for growth and spreading [5]. The rate of growth of bladder tumours is parallel to the presence of a rich blood supply formed by a large network of immature disturbed blood vessels in addition to the density of this network [6]. The intra-tumoural microvessel density (MVD), which is a measure of tumour angiogenesis in bladder cancer, was proved to correlate with tumoural malignancy [7], and its increase is associated with poorer a survival rate and increased rates of recurrence and mitotic index [8]. On the other hand, magnetic resonance imaging (MRI) has been shown to be superior to other imaging modalities in the diagnosis of bladder neoplasms, especially of superficial and multiple tumours and in tumour staging due to its high soft tissue discrimination [9]. Dynamic contrast-enhanced MRI (DCE-MRI) using fast MRI sequences obtained before, during, and after the rapid intravenous administration of gadolinium gives valuable data regarding both morphologic changes and physiological alteration associated with bladder malignancy [10]. The patterns of contrast enhancement on DCE-MRI are likely related to tumour angiogenesis. It could be used to assess the therapeutic response in patients who have not undergone surgical removal of their tumours [9]. We aimed at this study to assess the role of DCE-MRI and its correlation with tumour angiogenesis in the evaluation of urinary bladder cancer. Patients and Methods After approval of the Institutional Review Board of the Faculty of Medicine of Tanta University in June 2013, this study was performed at the Urology and Nephrology Center at Mansoura University between November 2013-June 2016, including a 1-year follow-up period. Ninety-four patients were initially included in the study, from whom 13 patients were excluded, t3 patients did not complete the study due to
claustrophobia, 4 patients were excluded due to bad image quality and motion artifacts, and 6 patients did not undergo biopsy, so we had no pathologic correlation for them. Finally, 81 patients were included in the study. Ethics committee approval, informed written consent, and explanation of the MRI procedure to the patients were completed. Privacy and confidentiality of all patient data were guaranteed, and there was a code number for every patient’s file that includes all investigations. All data provision were monitored and used for scientific purposes only. The cases were selected among the following: A. Patients with recent presumed diagnosis of bladder tumour at the outpatient clinic. In our country we have a high incidence of bladder cancer. We depend on the patient’s history of hematuria and lower urinary tract symptoms in suspect cases of bladder lesion. Then, we do outpatient cystoscopy for patients screening. B. Patients who came for follow-up after the management of histopathologically proven bladder cancer. All patients were subjected to the following: 1. Informed consent after full explanation of the benefits and risks of the procedure. 2. Full history taking: pelvic pain, painless gross hematuria, smoking, history of transurethral resection of bladder tumor (TURBT), and irritative bladder symptoms as dysuria, urgency, or frequency of urination. 3. Clinical examination: general signs of malignancy and metastasis, local examination by digital rectal or pervaginal examination to assess the mass, and its mobility or fixation to the pelvic wall. 4. Laboratory investigation: serum creatinine and urine cytology. 5. DCE-MRI within 3 weeks of initial diagnosis. The radiologic results then correlated with the histopathologic results. 6. Conventional cystoscopy. 7. Histopathologic investigation. Exclusion criteria included patients with renal failure, impaired renal function with glomerular filtration rate < 30 mL/min/1.73 m2 due to the risk of nephrogenic systemic fibrosis; past history of allergy from the MRI contrast media; artificial metallic prosthesis, eg, ‘‘artificial valves or cardiac pacemakers,’’ patients who refuse the study and those with severe claustrophobia.
DCE-MRI as diagnostic tool / Canadian Association of Radiologists Journal xx (2019) 1e10
MRI Protocol Patients were instructed to begin drinking water 1 hour before the MRI study and to present with a full bladder. In 8 patients with a urethral catheter, 250e400 mL sterile saline was used to distend the bladder. During the imaging procedure, fullness of the bladder was checked at localizer images and the examination was delayed if the bladder was not full. All MRI studies were done on a 1.5 Tesla Signa Horizon LX Echospeed scanner (GE Medical Systems, Milwaukee, WI) using a phased array pelvic coil. The procedure began by obtaining a localizer (scout image) to identify the pelvis. The following sequences were performed: A. T2-weighted high resolution sequences of the pelvis in axial, sagittal, and coronal directions to locate the lesions and to reveal their morphologic characteristics using the following parameters: time to repetition ¼ 7000e8000 ms, time to echo ¼ 90e 102 ms, band width ¼ 20e83 kHz, 256 256 matrix, slice thickness of 3 mm, intersection gap of 0.5 mm, and a field of view ¼ 20 20 cm. B. DCE-MRI: It was performed using fast spoiled gradient T1 sequence (pre- and post-contrast) in the axial, coronal, and sagittal planes by using the following parameters: repetition time ¼ 68 ms, time to echo ¼ 2 ms, 5 mm thickness, no inter-slice gap, flip angle 60 degrees, field of view 32 32 cm, and matrix 256 160. Images obtained before and immediately after manually injected gadolinium (gadoteric acid [Dotarem 0.5 mmoL/ mL]) at a dose of 0.1 mmol/kg of body weight (maximum, 20 mL) intravenously through the antecubital vein, followed immediately by a flush of 20 mL of saline solution. We tried to be fast as possible through a wide bore cannula to keep time less than 10 seconds for contrast and the same time for saline. Twelve sequential gradient T1 series, each lasting 15 seconds, were performed in 3 minutes.
3
hypointensity and enhances late. We used the following staging system according to Takeuchi et al [11]. Stage T1 Intact muscle layer at the base of the tumour that shows low signal intensity on T2-weighted MRI and no early enhancement on DCE-MRI. Stage T2 Disrupted bladder wall muscle’s hypointense line and early enhancement of the lesion without perivesical fat infiltration. Stage T3 Lesion with an irregular shaggy outer border and streaky areas in perivesical fat of the same signal intensity as the tumour. Stage T4 Lesion extending into an adjacent organ or abdominal and pelvic side walls. TimeeSI Curves TimeeSI curves were also constructed from signal intensity values obtained from freely drawn regions of interest selected on the basis of optimal visualization of the lesion and the region of greatest enhancement [12]. Type I (Ascending) Curve Time-SI curve shows enhancement, followed by a slow increase. Type II (Plateau) Curve
Image Post Processing Time-SI curve shows enhancement, followed by a plateau. The data were processed by using the software installed on the workstation system provided by the machine vendor (Advantage Workstation, version 4.6 [GE Healthcare, Bue, France]). The region of interest (ROI) for MRI enhancement was obtained from the region of maximum enhancement in the tumour that correlate with the areas of high vascularity with generation of time-signal intensity (SI) curves. The sizes of the ROIs were between 1-2 cm2. Three ROIs were placed within the masses for SI calculation, and we chose the highest value.
Type III (Descending) Curve Time-SI curve shows enhancement, followed by washout. Final diagnosis was confirmed on histopathology of the operated specimen. Staging and grading of tumours were analyzed on MRI and compared with the operative and histopathologic findings. Analysis of time-SI curves was done in all cases. Histopathologic Analysis
Image Analysis With DCE-MRI, bladder tumours, mucosa, and submucosa enhance early, but the muscle layer maintains its
Routinely processed, paraffin-embedded bladder tissues were used in this study. Three to 4 microns thick serial sections mounted on coated slides with Histogrip solution
4
O. A. Hassanien et al. / Canadian Association of Radiologists Journal xx (2019) 1e10
(Zymed Laboratories, Inc, San Francisco, CA), and subjected to: (1) haematoxylin and eosin stain for routine histopathologic examination, typing, grading, and staging of tumour; (2) immunohistochemical ‘‘IHC’’ staining for localization and evaluation of CD34 immunoreactivity. The CD34 is an endothelial cell marker, Ab-1 (Clone QBEnd/10). Mouse monoclonal antibody. (NeoMarkers, Inc) to detect the MVD. The MVD, defined as capillaries and small venules, was assessed in areas with solid tumour morphology away from any artifact or necrosis and without prior knowledge of patient outcome. After immuno-histochemical staining with CD34, the vessel counting was performed in 4 areas of maximal neovascularization where the highest number of discrete microvessels was stained (hot spots). Low power light microscopy at 40 magnification was used to scan the often heterogeneous tumour sections for these areas. At 250 magnification, counts were made of all distinct brown staining endothelial cells or cell clumps. The MVD was defined as the mean of the highest 4 counts. On the basis of the MRI findings, specimen sectioning for pathologic examination was performed without knowledge of patients’ outcome by neither the pathologist nor the radiologist. Statistical Data Data were analyzed using SPSS version 21.0 (IBM). Diagnostic efficacy was depicted in terms of sensitivity, specificity, and overall accuracy. The level of confidence was kept at 95%, hence a P < .05 indicated a significant association. Sensitivity, specificity, and accuracy of DCE-MRI were assessed on a stage-by-stage basis, and the gold standard was the histopathologic confirmation in all cases. We used the following statistical methods: A. Distribution of data was analyzed by mean and standard deviation. B. Frequency tables were analyzed using c2 test. C. Differences in between the groups were analyzed using independent-samples tests. D. Reported P values were obtained by using two-sided tests, and in all analysis, P < .05 was regarded as the limit of statistical significance. E. Receiver operating characteristic (ROC) curves were obtained for measuring the sensitivity and specificity of different variables. F. Kappa statistics were calculated to determine the strength of agreement and the harmony between the DCE-MRI stage and the histopathologic. stage. Value of kappa (< 0.20: poor), (0.21e0.40: fair), (0.41e0.60: moderate), (0.61e0.80: good), and (0.81e1.00: almost perfect). G. Spearman’s correlation test was used to determine the strength of the relationships between DCE-MRI results, MVD, and histopathologic and stage and grade (r ¼ 0.0e0.2 ¼ No or very low correlation),
Table 1 Number and classification of the 81 cases according to their main complaint Number of Cases Complaint Hematuria Follow-up post TURBT Irritative bladder symptoms Incidental Total
Benign 3 (30%) 3 (30%) 4 (40%) 10 (100%)
Malignant 58 9 2 2 71
(81.7%) (12.7%) (2.8%) (2.8%) (100%)
TURBT ¼ transurethral resection of bladder tumor.
(r ¼ 0.2e0.4 ¼ low correlation), (r ¼ 0.4e 0.6 ¼ reasonable correlation), (r ¼ 0.6e0.8 ¼ high correlation), and (r ¼ 0.8e1.0 ¼ very high correlation).
Results The current study included 81 patients (70 male [86.4%] and 11 female [13.6%]). The age ranged between 20-80 years, with mean age 59.25 years 10.167 standard deviation (SD). Seventy-one cases (87.7%) were pathologically proven to be malignant, while the other 10 patients (12.3%) were proved to be benign. Hematuria was the main complain, as it was present in 61 patients (75.3%), while 12 patients (14.8%) came for regular follow-up post TURBT, and 6 patients (7.4%) had irritative bladder symptoms (Table 1). Regarding the 71 malignant tumours, they were pathologically classified according to T staging: T1: 15 cases (21.1%), T2: 22 cases (31%), T3: 32 cases (45.1%), T4: 2 cases (2.8%); one of them was T4a, but was inoperable (the patient was not fit), and the other was T4b. Both were managed by taking chemotherapy, and there was no pathologic staging for them, so staging depended on MRI only in these 2 cases.
MR T2 W SI of the Bladder Lesions Ten cases were pathologically proven to be benign in nature: 6 cases (60%) had intermediate SI, 3 cases (30%) had hypointense SI, and 1 case (10%) had hyperintense SI at T2 W. Seventy-one cases were pathologically proven to be malignant in nature. The most common SI was intermediate: 49 cases (69%), 19 cases (26.8%), had hypointense SI, and 3 cases (4.2%) had hyperintense SI at T2 W (Table 2). Table 2 Classification of the cases according to MR T2 W SI of the bladder lesions T2 W SI
Benign
Malignant
Intermediate Hypointense Hyperintense
6 (60%) 3 (30%) 1 (10%)
49 (69%) 19 (26.8%) 3 (4.2%)
MR ¼ magnetic resonance; SI ¼ signal intensity.
DCE-MRI as diagnostic tool / Canadian Association of Radiologists Journal xx (2019) 1e10
5
Table 3 Type of SI curve in correlating with malignant pathologic staging Type of the curve Pathologic staging
Type I (Ascending)
Type II (Plateau)
Type III (Descending)
Total
T1 T2 T3 T4 (MR staging) Total
4 2 1 1 8
11 10 1 22
10 30 1 41
15 22 32 2 71
(26.7%) (9.1%) (3.1%) (50%) (11%)
(73.3%) (45.5%) (3.1%) (31%)
(45.5%) (93.8%) (50%) (58%)
(100%) (100%) (100%) (100%) (100%)
MR ¼ magnetic resonance; SI ¼ signal intensity.
Figure 1. 31-year-old female patient, presented with irritative bladder symptoms and frequent attacks of hematuria 1 year ago, she had no history of previous transurethral resection of bladder tumor. Proved to be follicular cystitis. Magnetic resonance imaging T2 high resolution: axial (A), coronal (B), and sagittal (C) views revealed: reduced bladder capacity, irregular circumferential bladder wall thickening with sessile lesions at the bladder dome, anterior, and right lateral walls, displaying hypointense signal intensity with infiltration of the perivesical fat (red arrows). Dynamic contrast-enhanced magnetic resonance imaging, axial view. (D) Early post-contrast gradient T1 showed mild diffuse enhancing circumferential bladder thickening (yellow arrow). (E) Late post-contrast gradient T1: persistent and increasing enhancement (yellow arrow). (F) Timeesignal intensity curve shows slow enhancement followed by slow increase with no washout. (G) Haematoxylin-eosin stained slide: follicular cystitis. (H) CD34 expression (250). Mean of MVD ¼ 84.
6
O. A. Hassanien et al. / Canadian Association of Radiologists Journal xx (2019) 1e10
Figure 2. 57-year-old male patient with cancer bladder, presented with hematuria 4 months ago, he had no previous history of transurethral resection of bladder tumor. Magnetic resonance imaging T2 high resolution: axial (A), coronal (B), and sagittal (C) views revealed: intra-vesical fungating soft tissue mass, displays intermediate signal intensity (SI), arising from the left lateral bladder wall (red arrow), encroaching upon the left ureteric orifice with intact hypointense SI of the related muscle layer. Dynamic contrast-enhanced magnetic resonance imaging revealed, axial view. (D): Early post-contrast gradient T1 showed an intra-vesical soft tissue mass arising from the left lateral bladder wall, encroaching upon the left ureteric orifice. The mass showed earlier enhancement than the muscle layer (yellow arrow). (E) Late post-contrast gradient T1 sequence: The detrusor muscle is seen intact and showed delayed enhancement (yellow arrow). Persistent tumour enhancement is also noted. No extra-vesical fat infiltration. (F) TimeeSI curve shows rapid enhancement followed by plateau. Stage T1. (G) Haematoxylin-eosin stained slide showed grade II papillary TCC. Stage T1. (H) CD34 expression (250). Mean of MVD ¼ 37.
Types of TimeeSI Curve for the 81 Patients Included in the Study DCE-MRI results according to the washout slope of the time-SI curve revealed: 17 cases (21%) showed ascending curve (type I), 23 cases (28%) showed plateau curve (type II), and 41 cases (51%) showed descending curve (type III). In the 71 cases pathologically proven to be malignant, the type of the time-SI curve was as follows: 41 cases (58%) showed type III (descending) curve, 22 cases (31%) showed type II (plateau), and 8 cases (11%) showed type I
(ascending) curve (Table 3). Ten cases proved to be cystitis: 9 cases (90%) showed type I (ascending) curve and 1 case (10%) showed type II (plateau) curve (Figure 1).
Type of Curve in Correlation With Pathologic Tumour Staging Fifteen cases proved to be stage T1: 4 cases (26.7%) showed type I (ascending) curve and 11 cases (73.3%) showed type II (plateau) curve (Figure 2).
DCE-MRI as diagnostic tool / Canadian Association of Radiologists Journal xx (2019) 1e10
7
Figure 3. 52-year-old male patient with cancer bladder, presented with hematuria 6 months ago; he had no previous history of transurethral resection of bladder tumor. Magnetic resonance imaging T2 high resolution: axial (A), coronal (B), and sagittal (C) views revealed: intra-vesical fungating soft tissue mass, displays intermediate signal intensity (SI), arising from the left postero-lateral bladder wall (red arrows), encroaching upon the bladder neck and left ureteric orifice with loss of hypointense SI of the related muscle layer. Dynamic contrast-enhanced magnetic resonance imaging revealed: axial (D) early post-contrast gradient T1 showed earlier enhancement of the mass than the muscle layer (yellow arrow). (E) Late post-contrast gradient T1 sequence: the detrusor muscle showed delayed enhancement as well as interruption by the tumour (yellow arrow). No extra-vesical fat infiltration. (F) TimeeSI curve showed rapid enhancement followed by washout. Stage T2. (G) haematoxylin-eosin stained slide showed grade III papillary TCC. Stage T2. (H) CD34 expression (250): mean of MVD ¼ 68.
Twenty-two cases proved to be stage T2: 2 cases (9.1%) showed type I (ascending) curve, 10 cases (45.5%) type II (plateau) curve, and 10 cases showed (45.5%) type III (descending) curve (Figure 3). Thirty-two cases proved to be stage T3: 1 case (3.1%) showed type I (ascending) curve, 1 case (3.1%) showed type II (plateau) curve, and 30 cases (93.8%) showed type III (descending) curve. Two cases proved to be stage T4: 1 case (50%) showed type I (ascending) curve, and 1 case (50%) showed type III (descending) curve (Table 3). The sensitivity of dynamic MRI in stage T1 bladder tumour was 80%; in stage T2, it was (90.9%); and in stage T3, it was (96.9%). In T4 bladder tumours, one was stage T4a
(inoperable, as the patient was unfit for operation), and the other was T4b, so management depended upon MRI staging, by taking chemotherapy for both cases. The overall accuracy of DCE-MRI in tumour staging was 89.5%. Over-staging occurred in 6 cases (7.4%): 3 cases T1, 2 cases T2, and 1 case (cystitis). Under-staging occurred in 1 case, which pathologically diagnosed as T3, and we misdiagnosed it as cystitis (Table 4). Kappa Statistics To determine the level of agreement and the harmony between the DCE-MRI stage and the histo-pathologic stage,
8
O. A. Hassanien et al. / Canadian Association of Radiologists Journal xx (2019) 1e10
Table 4 Sensitivity values of dynamic MRI compared with pathologic staging Number of cases detected by dynamic MRI/total number detected by pathology
Pathologic Staging
Sensitivity
T1 12/15 80%
T2 20/22 90.9%
T3 31/32 96.9%
Cystitis 9/10 90%
Overall accuracy
P value
k statistics
Spearman correlation test
89.5%
P ¼ .001
k ¼ 0.880
r ¼ 0.965
P < .05 was regarded as the limit of statistical significance. Kappa statistics: k ¼ 0.81e1.00 is almost perfect. Spearman’s correlation test: r ¼ 0.8e1.0 ¼ very high correlation. MRI ¼ magnetic resonance imaging.
DCE-MRI agreement was almost perfect in assigning the tumour stage (k ¼ 0.880, P ¼ .001).
staging (r ¼ 0.606, P ¼ .001) where (r ¼ 0.6e0.8 ¼ high correlation).
Correlation Between the DCE-MRI Staging and Histopathologic
Correlation Between Dynamic MRI Curves and MVDs
Staging (Spearman correlation test) A very high positive correlation was noted between the DCE-MRI staging and histopathologic staging (r ¼ 0.965) where r ¼ 0.8e1.0 ¼ very high correlation (Table 4). Correlation Between the Type of SI Curve and Histopathologic Staging High positive correlation between the type of curve and pathologic staging (r ¼ 0.646, P ¼ .001) where r ¼ 0.6e0.8 ¼ high correlation. Tumour Angiogenesis
Spearman Correlation test showed reasonable positive correlation between the washout slope of dynamic MRI curve and mean MVD (r ¼ 0.471, P ¼ .001), where r ¼ 0.4e0.6 is reasonable correlation and P ¼ .001 is statistically highly significant. Discussion The formation of new blood vessels from pre-existing vasculature (angiogenesis) is a critical part of tumour growth and progression [13,14]. Tumour growth is affected by inhibition of angiogenesis, and this may even result in tumour regression [15]. The majority of published studies have shown a positive correlation between intra-tumoural
Microvessel densities were found to be between 17-122 vessels/area (mean ¼ 55 23.3 SD vessels/area). The mean of MVD in the 10 benign cases was 77.63 16.73 SD. The mean of MVD of the 71 malignant cases was 53.77 22.36 SD. Values more than the cutoff value ¼ 76.13 MVD are cystitis. The area under curve is 0.863 (95% confidence interval 0.723e1.000). Sensitivity is 90%. Specificity is 91%. P value is .001, which is statistically highly significant (Figure 4). MVD Correlating With Pathologic Tumour Staging Regarding the tumour staging, in the 15 cases were diagnosed as T1, the mean of the MVD was 27.12 7.14 SD. In the 22 cases diagnosed as T2, the mean of MVD was 50.13 16.78 SD. In 32 cases diagnosed as T3, the mean of MVD was 65.72 19.65 SD. In 2 cases (staged by MRI) as T4, the mean of MVD was in case diagnosed as T4a was 72, and in the other case diagnosed as T4b, mean of MVD was 21 (Table 5). Spearman correlation test was done to determine the strength of the relationship between the mean MVD and tumour staging: There is a high positive correlation was also noted between the mean MVD and tumour pathologic
Figure 4. Receiver operating characteristic (ROC) curve: sensitivity and specificity of microvessel density in differentiation between type of cells (benign and malignant cells).
DCE-MRI as diagnostic tool / Canadian Association of Radiologists Journal xx (2019) 1e10 Table 5 Mean microvessel density correlating with pathologic tumour staging Pathologic staging
Number
Mean of MVD
SD
T1 T2 T3
15 22 32
27.12 50.13 65.72
7.14 16.78 19.65
MVD ¼ microvessel density; SD ¼ standard deviation.
MVD, ‘‘a measure of tumour angiogenesis’’ and prognosis in bladder tumours [7,16e18]. Radiological evaluation is a significant part of diagnosis and staging of bladder cancer. Functional dynamic contrast-enhanced MRI can assess the microcirculation and visualize the neo-angiogenesis of malignant tissues via the dynamic signal enhancement of a contrast agent [19,20]. This study was carried out with 81 patients (70 men and 11 women); 71 cases were proven to be malignant. Mean age was 60 years; this copes with the recorded age and sex incidence of bladder tumours [21,22]. In the current study, the sensitivity of dynamic MRI in in the radiological staging compared to histopathologic staging in stage T1 bladder tumour was 80%. In stage T2, it was 90.9%, and in stage T3, was (96.9%), with overall accuracy 89.5%. Two cases with T4 bladder tumours were present in this study; one was stage T4a (inoperable, as the patient was unfit for operation), and the other was T4b, so management depended upon MRI staging, by taking chemotherapy for both cases. Our study showed higher sensitivity than that of Gupta et al [23]. In their study, the sensitivities of dynamic MRI in stage (T1, T2, T3, T4) bladder tumour were 62.5%, 83.3%, 50%, and 100%, respectively. Over-staging was seen in 6 patients (7.4%) in this study, compared to study by Gupta et al [23] where over-staging occurred in 20% of cases. The reason of overestimation could be due to abnormal signals in perivesical fat, especially if associated with cystitis that could be incorrectly identified as tumour invasion. Vascular hyperemia adjacent to the tumour could cause these abnormal signals. On the other hand, under-staging occurred in 1 case, which pathologically diagnosed as T3, and we misdiagnosed it as cystitis. This case has previous transurethral resection before reference to our center, and the mass could not be visualized among the thickened bladder wall. In our study, statistically, the extent of agreement between dynamic contrast MRI, and histopathologic staging was almost perfect (k ¼ 0.880) and highly significant (P ¼ .001) with positive very high correlation (r ¼ 0.965). These results are higher compared to the study done by both of Gupta et al [23] on 60 patients that resulted in good agreement equals (k ¼ 0.690) and significant (P < .001), and with Tuncbilek et al [9] on 24 patients that resulted in fair agreement (k ¼ 0.365) and significant (P ¼ .002). It must be emphasized that MVD evaluation under the microscope is somewhat subjective and inter-observer variation can be substantia [7]. So, we proceeded for the data
9
analysis of the angiogenesis of the bladder lesions and correlating them with the type of cells, tumour stage, grade, and the incidence of recurrence. We studied the correlation between the MVD and type of cells ‘‘benign and malignant’’; we noticed the higher vascularity of cases of cystitis, compared to the vascularity of malignant tumours. This could be explained by the higher vascularity of the active cells, edema, inflammatory infiltrate, and vascular congestion during the inflammatory process. To our knowledge, our study was the first to correlate the mean of MVD with the type of cells (cystitis vs malignant). From the ROC curve, by applying the cutoff value ¼ 76.13 MVD, the higher MVD above the cutoff value are cystitis with sensitivity (90%), specificity (91%). P value (.001), which is statistically highly significant. We studied the correlation between the MVD and pathologic tumour staging that revealed high positive correlation between MVD and tumour stage (r ¼ 0.606, P ¼ .001). In the current study, angiogenesis increases in parallel with increasing tumour stage and grade, apart from T4 (2 cases), due to small group number; also, one of them was diagnosed as undifferentiated cell carcinoma with large areas of cystic necrosis, so it could explain the lower MVD than the T2 and T3 tumours. This is fully consonant with the observations by Gehani et al [7] and Canoglu et al [24] that also confirmed that MVD in bladder carcinoma correlates with the tumour grade, stage, and its malignant potential. Our study revealed that there is a reasonable positive correlation was also noted between the washout slope of dynamic MRI curve and mean MVD (r ¼ 0.471, P ¼ .001) that copes with the study done by Tuncbilek et al [9]. The limitations of our study were related to the technique of dynamic MRI using the manual technique of injection, which may affect the results; also, the short follow-up period of patients. At the end of this study, we can conclude that DCE-MRI supplemented with time-SI curve is a promising non-invasive diagnostic tool in the staging of cancer bladder through the tumoural vascular behavior and angiogensis. This may allow DCE-MRI to be used in preoperative predictions of tumour behavior and biologic activity that can affect the therapy planning with antiangiogenetic therapy. We recommend a larger and longer study to confirm the findings. References [1] Gang J, Fan J, Wang Q, et al. Decreased REG1 a expression suppresses growth, invasion and angiogenesis of bladder cancer. Eur J Surg Oncol 2017;43:837e46. [2] Zarzour AH, Selim M, Abd-Elsayed AA, et al. Muscle invasive bladder cancer in Upper Egypt: the shift in risk factors and tumor characteristics. BMC Cancer 2008;8:250. [3] Felix AS, Soliman AS, Khaled H, et al. The changing patterns of bladder cancer in Egypt over the past 26 years. Cancer Causes Control 2008;19: 421e9. [4] Salem HK, Mahfouz S. Changing patterns (age, incidence, and pathologic types) of schistosoma-associated bladder cancer in Egypt in the past decade. Urology 2012;79:379e83.
10
O. A. Hassanien et al. / Canadian Association of Radiologists Journal xx (2019) 1e10
[5] T€ urkbey B, Thomasson D, Pang Y, et al. The role of dynamic contrastenhanced MRI in cancer diagnosis and treatment. Diagn Interv Radiol 2010;16:186e92. [6] Xie Q, Zhang J, Wu P-H, et al. Bladder transitional cell carcinoma: correlation of contrast enhancement on computed tomography with histological grade and tumour angiogenesis. Clin Radiol 2005;60:215e23. [7] Gehani K, Al-Kikhia L, Mansuri N, et al. Angiogenesis in urinary bladder carcinoma as defined by microvessel density (MVD) after immunohistochemical staining for Factor VIII and CD31. Libyan J Med 2011;6. [8] Hall A. The role of angiogenesis in cancer. Compar Clin Pathol 2005; 13:95e9. [9] Tuncbilek N, Kaplan M, Altaner S, et al. Value of dynamic contrastenhanced MRI and correlation with tumor angiogenesis in bladder cancer. AJR Am J Roentgenol 2009;192:949e55. [10] Choyke PL, Dwyer AJ, Knopp MV. Functional tumor imaging with dynamic contrast-enhanced magnetic resonance imaging. J Magn Reson Imaging 2003;17:509e20. [11] Takeuchi M, Sasaki S, Ito M, et al. Urinary bladder cancer: diffusionweighted MR imaging-accuracy for diagnosing T stage and estimating histologic grade. Radiology 2009;251:112e21. [12] Cheng L, Li X. Breast magnetic resonance imaging: kinetic curve assessment. Gland Surg 2013;2:50e3. [13] Iagaru A, Gambhir SS. Imaging tumor angiogenesis: the road to clinical utility. AJR Am J Roentgenol 2013;201:183e91. [14] Folkman J. Role of angiogenesis in tumour growth and metastasis. Semin Oncol 2002;29:15e8. [15] Turkbey B, Kobayashi H, Ogawa M, et al. Imaging of tumor angiogenesis: functional or targeted? AJR Am J Roentgenol 2009;193: 304e13.
[16] Chikazawa M, Inoue K, Fukata S, et al. Expression of angiogenesisrelated genes regulates different steps in the process of tumor growth and metastasis in human urothelial cell carcinoma of the urinary bladder. Pathobiology 2008;75:335e45. [17] Sarı A, Ermete M, Dag F, et al. Evaluation of angiogenesis in urothelial carcinoma of bladder and its relation with prognostic factors. Ege Journal of Medicine 2007;46:135e40. [18] Goddard JC, Sutton CD, Furness PN, et al. Microvessel density at presentation predicts subsequent muscle invasion in superficial bladder cancer. Clin Cancer Res 2003;9:2583e6. [19] Tekes A, Kamel I, Imam K, et al. Dynamic MRI of bladder cancer: evaluation of staging accuracy. AJR Am J Roentgenol 2005;184: 121e7. [20] Rose T, Lotan Y. Advancements in optical techniques and imaging in the diagnosis and management of bladder cancer. Urol Oncol 2018;36: 97e102. [21] Epstein JI, Lotan TL. The lower urinary tract and male genital system. In: Kumar V, Abbas A, Aster JC, editors. Robbins & Cotran Pathologic Basis of Disease. 9th ed. Toronto: Saunders Elsevier; 2015. p. 959e90. [22] Rabie E, Faeghi F, Izadpanahi MH, et al. Role of dynamic contrastenhanced magnetic resonance imaging in staging of bladder cancer. J Clin Diagn Res 2016;10:TC01e5. [23] Gupta N, Sureka B, Kumar MM, et al. Comparison of dynamic contrast-enhanced and diffusion weighted magnetic resonance image in staging and grading of carcinoma bladder with histopathological correlation. Urol Ann 2015;7:199e204. [24] Canoglu A, G€og€us¸ C uk Y, et al. Microvessel density as a prog¸ , Bed€ nostic marker in bladder carcinoma: correlation with tumor grade, stage and prognosis. Int Urol Nephrol 2004;36:401e5.