Introduction

Bladder cancer is the most common urothelial malignancy [1, 2]. It is the ninth most common cancer overall with an estimated 80,470 new cases (61,700 among males and 18,770 among females) and 17,670 deaths (12,870 among males and 4,800 among females) in the US in 2019 [1, 2]. Men are approximately three times more likely than women to be diagnosed with bladder cancer; patients are usually diagnosed in the 8th decade of life [3]. Painless hematuria and frequency of urination are the most common symptoms; urinary tract obstruction and pain are limited to advanced cases. Patients with symptoms of bladder cancer are typically evaluated with computed tomography (CT) urography and cystoscopy, and when a suspected lesion is found, transurethral resection of the bladder tumor (TURBT) to both verify the diagnosis and stage the disease. Clinical management and prognosis depend in large part on the presence of disease within bladder muscle—muscle-invasive bladder cancer (MIBC) versus superficial, non-muscle-invasive bladder cancer (non-MIBC)—and the presence of nodal and metastatic lesions [2, 3].

Approximately 75–85% of patients with bladder cancer present with non-muscle-invasive disease (Ta, T1 or Tis), which is typically treated with bladder-sparing techniques such as TURBT alone with or without adjuvant intravesical or systemic therapies [4]. MIBC (≥ T2 disease) is commonly treated with partial or radical cystectomy with pelvic lymph node dissection, neoadjuvant or adjuvant chemotherapy and radiation [2, 3, 5]. The detection of MIBC is important due to significant differences in outcome and treatment strategies. Currently, TURBT is the method of choice to detect muscle invasion, however, several studies have indicated that this method may understage the disease in up to 40% of patients [6, 7]. Therefore, a noninvasive alternative for accurate staging of the bladder cancer would be of immense value.

The Tumor, Node, Metastasis (TNM) system is currently the standard staging system used by the American Joint Committee on Cancer (AJCC) for bladder cancer (Table 1) [8]. Given the suboptimal performance of TURBT in local staging of the bladder cancer, several noninvasive morphological and functional imaging-based methods have been proposed that can improve the accuracy of local staging of the bladder cancer [9]. Nodal staging with conventional cross-sectional imaging [CT and magnetic resonance imaging (MRI)] is based on lymph node size, which often leads to understaging of metastatic disease to locoregional lymph nodes in up to 30% [10]. This review highlights the advantages and disadvantages of preoperative imaging studies for locoregional staging of bladder cancer, including the local staging and adjacent nodal involvement.

Table 1 Bladder cancer TNM staging system adapted from the American Joint Committee on Cancer (AJCC) 8th edition [8]
Full size table

Imaging modalities

Computed tomography (CT)

CT urography is the most widely used imaging modality for detection and staging of urothelial carcinoma with a reported diagnostic accuracy of 91% in detection and 35–55% in staging [11, 12]. At our institution, CT urography is performed as a three-phase examination that includes non-contrast, urothelial and excretory phases. Some institutions use a split-bolus technique, in which images in the urothelial and excretory phases are obtained at the same time with different injection times. This two-phase examination reduces scanning time and radiation dose, however, there is some evidence that contrast material used in the excretory phase can mask small lesions and limit the evaluation of tumor enhancement during a dedicated urothelial phase [13]. In a retrospective evaluation of 2600 patients with hematuria, Sadow et al. showed that CT urography can be used to detect bladder cancer; while the negative predictive value was 98%, the sensitivity was 79% [12] (Fig. 1). Although CT has restrictions in accurately differentiating between T1, T2 and T3a due to its inability to individually identify layers of the bladder wall, it may provide useful staging information regarding perivesical invasion of high-grade T3b and T4 tumors [13, 14] (Fig. 2). Several groups have investigated the role of CT in local staging of bladder cancer, for example Caterino et al. utilized three different CT techniques; imaging the bladder filled with urine, intravenous contrast material or air. In their study, patients were assigned to four groups; Ta–T1 (group I), T2–T3a (group II), T3b (group III) and T4 (group IV). Their results showed that the air-insufflated bladder technique had the highest accuracy in evaluating the degree of mural invasion in the first three groups. The accuracy of all techniques were the same (100%) in characterizing advanced tumors with adjacent organ invasion (T4) [15]. We summarize the results of CT-based studies for local staging of bladder cancer in Table 2; Overall, CT is modestly accurate in the detection of perivesicular invasion (≥ T3), ranging from 49 to 93%.

Fig. 1
figure 1

75-year-old woman with microscopic hematuria. At CT urography, a 1.9 cm papillary mass arises from the left bladder wall (arrow). The uninterrupted adjacent wall and lack of findings in the perivesical fat suggests the tumor is organ confined (≤ T3). These findings were concordant with results from TURBT

Full size image
Fig. 2
figure 2

67-year-old man with gross hematuria. CT urogram demonstrates a 7.8 cm mass arising from the left lateral bladder wall. Perivesical stranding and nodularity suggests extravesicular extension of tumor (≥ T3) (arrow). Large diverticulum arises from the right posterolateral bladder wall (arrowhead). Extravesicular extension of tumor was confirmed at surgery, with involvement of the rectosigmoid (T4)

Full size image
Table 2 Sensitivity, specificity and accuracy for detection of perivesicular invasion (≥ T3) and lymph node involvement with CT
Full size table

Identifying lymph node involvement on CT is usually based on lymph node size and shape. Lymph nodes are considered abnormal on imaging if the short axis diameter is greater than 8 and 10 mm in pelvis and abdomen, respectively [16]. This may lead to missing metastatic disease in lymph nodes smaller than 8 mm in short axis diameter, while over-diagnosing the metastatic involvement in reactive lymph nodes larger than 10 mm in short axis diameter in up to 30% of the patients [10]. Pichler et al. demonstrated that the best cutoff value for metastatic pelvic lymph node identification was 8 mm (sensitivity and specificity of 45.5% and 91.5%), Li et al. suggested that 6.8 mm is the optimal short axis diameter cutoff (area under the ROC curve: 0.815) [16, 17]. These studies showed that the sensitivity of CT images in detecting nodal metastases is strictly related to the size of the involved lymph node and reducing the size threshold can cause a high false-negative results. Lymph node size is important both for the diagnose of metastatic lymph nodes, and as an indicator of prognosis in patients with bladder cancer. In a study of 206 patients, Schmid et al. reported that when metastatic pelvic lymph nodes are greater than 5 mm in diameter, the risk of death due to bladder cancer significantly increases [18]. We summarize the results of CT-based studies in Table 2; overall, the performance of CT for diagnosis of lymph node metastasis is variable amongst different studies, ranging from 54 to 86%.

Even though CT cannot be used to precisely stage bladder cancer due to moderate accuracy in differentiating MIBC from superficial, non-MIBC and highly variable results for the diagnosis of metastatic lymph nodes, it remains the mainstay for initial staging of bladder cancer because of its cost-effectiveness, availability, and utility in detecting both metastatic disease in the entire abdomen and pelvis and when a CT urography protocol is used, synchronous upper tract cancers.

Magnetic resonance imaging (MRI)

Magnetic resonance (MR) urography provides valuable information on bladder and the synchronous upper urinary tract disease. In comparison to CT urography, MR urography has shown a lower sensitivity in detecting upper tract disease (67% vs 88–100%), however, it remains an alternative to CT urography in patients with contraindications to CT [32]. For local T staging of bladder cancer, MRI is superior to CT due to its high soft tissue contrast resolution [33]. The added value of MRI is in part due to the functional data derived from diffusion-weighted imaging (DWI) and dynamic contrast enhanced (DCE) imaging, which can provide additional information regarding tumor invasion, especially in differentiating superficial from muscle-invasive tumors [3, 34, 35]. Although new MRI sequences may improve our ability to further improve nodal staging, MRI and CT have been shown to perform similarly because nodal involvement is based on morphological patterns in both [3, 36]. Each individual sequence used when performing an MRI of the pelvis (or MR urography) is discussed with respect to both findings and contribution to staging.

T2-weighted imaging (T2WI)

T2WI is the mainstay of bladder cancer staging as it depicts individual layers of the bladder wall and their relationship to the tumor (Fig. 3). Superficial, non-MIBC appears as an intermediate to high-signal intensity mass on T2WI with an intact adjacent hypointense detrusor muscle. In contrast, muscle-invasive tumors interrupt the detrusor muscle [37] (Fig. 4).

Fig. 3
figure 3

81-year-old man with distant history of bladder cancer presents with hematuria. MR urogram demonstrates a 1.5 cm T2 hyperintense papillary mass with a stalk (arrow) (a and b). The mass demonstrates restricted diffusion on high b value DWI that was confirmed on ADC (c and d). Four-phase DCE images demonstrate avid enhancement with thickened subjacent inner layer of the bladder wall (arrow heads) (e–h). Surgery was consistent with high-grade T1 TCC

Full size image
Fig. 4
figure 4

75-year-old man with hematuria was found to have a polypoid bladder mass on MR urogram. The mass (arrow) at the posterior bladder wall demonstrates T2 hyperintensity on sagittal (a) and coronal (b) T2 weighted images. There is increased T2 signal within the subjacent bladder wall (arrowheads). However, there is no evidence of extension of the mass beyond the bladder wall. The mass demonstrates restricted diffusion on low and high b-value images (c and d). The mass is isointense to the bladder wall on precontrast T1-weighted images (e) and demonstrates enhancement after injection of contrast (f and g). Note enhancement deep into the muscular layer as demonstrated on post contrast images (curved arrow) consistent with pathologic staging of T2b

Full size image

Diffusion-weighted imaging (DWI)

DWI relies on detection of Brownian motion of free-water molecules. Tumors with high cell density typically lead to restricted diffusion of water molecules and show high-signal intensity on DWI. The apparent diffusion coefficient (ADC) map is a quantitative display of the degree of restricted water diffusion. An ADC value can also be derived; its value decreases as tumor aggressiveness increases [38, 39]. DWI has shown promise in both detection and staging bladder tumors as well as detecting lymph node involvement [40] (Fig. 5). In comparison to T2WI alone, studies have reported superior sensitivity, specificity and accuracy when both T2WI and DWI were used to differentiate superficial from invasive tumors [39, 40]. Wu et al. showed that the diagnostic performance of combined DWI and T2WI was more accurate than DWI alone, especially in differentiating Tis -T1 from ≥ T2 disease [41].

Fig. 5
figure 5

78-year-old man with painless hematuria. MR urogram T2WI (a) demonstrates a 6.0 cm mass along the right bladder base, with extension into a large right posterolateral bladder wall diverticulum (arrowhead). The mass demonstrates heterogenous enhancement (b) and marked restricted diffusion (c). Invasion of the perivesical (T3) adipose tissue (arrow) was demonstrated only on coronal images (d) and was confirmed at surgery

Full size image

In a study of 36 bladder cancer patients by Papalia et al., involved lymph nodes demonstrated diffusion restriction manifested as high-signal intensity on DWI with a b value of 1000 s/mm2 and low-signal intensity on ADC map. The ADC value of 0.86 × 10−3 mm3/s or less was considered to be malignant with sensitivity of 76.4% and specificity of 89.4% in this study [36].

Dynamic contrast-enhanced (DCE) imaging

DCE imaging is used to detect tumor based on tumor angiogenesis. Hypervascular malignant lesions exhibit early enhancement on DCE images [42,43,44] (Fig. 6). Several studies have shown relative accuracies of DCE imaging in differentiating superficial and muscle-invasive tumors as well as between organ-confined and non-organ-confined tumors, ranging from 85–97% to 82–84%, respectively [35, 42]. Gupta et al. showed that both DCE imaging and DWI performed similarly when staging locally advanced T4 tumors with sensitivity of 100%, and specificity and accuracy of more than 96%, however, the performance of both techniques was significantly reduced in tumors of lower stages. The diagnostic accuracy of DWI in differentiating organ-confined from non-organ-confined tumors was higher than DCE imaging in this study but results were better when both were used in combination. Therefore, the authors recommended a combination of DWI and DCE imaging for preoperative staging of bladder cancer [44]. This study and several others have demonstrated the superiority of multiparametric MRI (mpMRI), in which a combination of anatomic (T2WI and DCE) and functional imaging such as DWI, diffusion tensor imaging (DTI) and perfusion-weighted imaging (PWI) were used for staging bladder cancer [37]. Panebianco et al. demonstrated that a comprehensive mpMRI protocol with a combination of T2WI, PWI, DWI, and DTI outperformed other combinations of MRI sequences for the diagnosis of muscle-invasive bladder cancer (AUC: 0.992) [33]. We summarize the results of MRI-based studies in differentiating superficial versus muscle-invasive bladder cancer in Table 3 and organ-confined versus locally advanced tumors in Table 4.

Fig. 6
figure 6

62-year-old man with incidentally detected bladder mass on prostate MRI. The mass demonstrates T2 hyperintensity and is confined to the mucosal surface (a and b) with restricted diffusion as seen on high b-value image (arrow) (c). DCE images demonstrate avid enhancement of the mass with no interruption of the subjacent bladder wall. TURBT was performed and the mass was shown to be grade 1 TCC and staged as Ta. Staging CT demonstrated the mass (arrowhead), however, the distinction of the mass and the adjacent wall was impossible on CT (g)

Full size image
Table 3 MRI characteristics in differentiating of superficial (Ta, T1) from muscle-invasive bladder cancer (T2, T3, T4)
Full size table
Table 4 MRI characteristics in differentiating organ-confined (≤ T2) from non-organ-confined (≥ T3) bladder cancer
Full size table

Ultra-small superparamagnetic particles of iron oxide (USPIO) enhanced-MRI

There are several alternatives to conventional gadolinium-based contrast agents. One of these agents is USPIO [45]. When injected, USPIO is a lymphotropic iron oxide agent that is absorbed by normal macrophages within lymph nodes. Normal lymph nodes absorb the agent, and therefore, show a signal intensity loss of more than 15% on T2WI due to iron-induced magnetic susceptibility. Lymph nodes replace with tumor do not contain macrophages, therefore do not take up the agent and show no signal loss on the same sequence [45, 46]. In a study of 75 patients with either bladder or prostate cancer, preoperative evaluation of lymph nodes with (USPIO)-enhanced MRI showed sensitivity, specificity and accuracy of 58.3, 83 and 76.4, respectively [46]. Another study on the same population reported improved detection rate of metastatic lymph nodes on combined USPIO-DW-MRI with a median time spent of 9 min for reading pelvic MRI in comparison to 32 min with USPIO-MR alone since benign lymph nodes are not displayed on USPIO-DWI and the need for the node to node comparison between pre and post contrast sequences was obviated [47]. We summarize the results of studies on the role of MRI in differentiating metastatic lymph nodes in Table 5.

Table 5 Sensitivity, specificity and accuracy of specific MRI pulse sequences in detecting lymph node involvement in patients with bladder cancer
Full size table

Positron emission tomography (PET)

[18F] fluoro-d-glucose (18F-FDG) PET scan is the most common molecular imaging technique for preoperative staging of various malignancies, however, there are some limitations in the evaluation of bladder cancer. As an analog of glucose, 18 F-FDG is excreted in the urine, and therefore, accumulates in the bladder. As a result, the detection of bladder masses can be challenging when there is a high level of activity in the bladder (Fig. 7). Administration of oral hydration and forced diuresis by an intravenous diuretic, bladder catheterization with irrigation, and utilization of different radiotracers are methods applied to overcome this limitation [55,56,57]. Forced diuresis with oral hydration method has been shown to be the safest method to induce accelerated excretion of urinary radioactivity, providing high-resolution images [56].

Fig. 7
figure 7

74-year-old man with left bladder wall mass. There is disruption of the T2 dark detrusor muscle with extension into the perivesical fat (T3) on T2WI (a) and DCE (b) images (arrows). The mass demonstrates redistricted diffusion with high intensity on high b-value DWI image (c) and low signal on corresponding ADC map (d). PET-CT (e) demonstrates avid uptake of the radiotracer at the site of extravesical extension next to the left ureter (arrowhead). Extravesical extension is also seen on CT (f)

Full size image

In addition to the role of PET-CT in locoregional staging, this modality can help in establishing the prognosis. Drieskens et al. reported a median survival of 13.5 months in patients with positive nodal and distant metastasis on PET-CT scans, and 32 months when the scans were negative for metastasis using the Kaplan–Meier survival curves [58]. To evaluate the clinical utility of PET-CT in patients with bladder cancer, the treatment plan for muscle-invasive bladder cancer was compared before and after PET-CT imaging. Treatment was altered in 20–27% of cases who underwent preoperative PET-CT, mainly due to upstaging by this technique in comparison with conventional imaging [59, 60].

Whether PET-CT is better than CECT in nodal staging is unclear. Some suggest FDG PET-CT is superior with sensitivities ranging from 36 to 78% for PET-CT versus 9.1–44% for CECT [21, 24, 26], others have not [16, 22, 29, 61]. Given the suboptimal performance of PET-CT in the diagnosis of small metastatic lymph nodes, Goodfellow et al. recommended using PET-CT only in patients with either enlarged pelvic LNs and a small primary bladder tumor, in patients with extra pelvic nodal metastases, and in patients who have indeterminate lesions suspicious for metastasis [27]. Although PET-CT showed higher sensitivity for pelvic lymph node detection (68%) compared to PET (46%) or CT (46%) alone, they did not suggest the routine use of PET-CT in the preoperative staging of patients with bladder cancer due to its limited additional value and the high cost.

Non-FDG PET tracers, such as radiolabeled 11C-choline and 11C-acetate have been used to preoperatively assess lymph node involvement of bladder cancer. 11C-acetate and 11C-choline both show increased uptake in neoplastic lesions and since little to no tracer is excreted in the urine, the primary tumor can be seen [62]. Radiolabeled 11C-choline is thought to provide better definition between the tumor and the background compared to acetate, mostly due to the lower background of choline than acetate and not higher uptake of choline [63]. In a study by Picchio et al., 11 C-choline PET demonstrated fewer false positive lymph nodes than did CT [64]. In comparison to 11 C-choline PET-CT for bladder cancer staging, 18F-FDG showed a higher positive predictive value in detecting extravesical lesions (88.2% vs. 79.4%) [65].

The standardized uptake value (SUV) is used to quantitatively assess FDG uptake relative to the background. No specific SUV can be used to differentiate malignant from benign lesions with complete accuracy [16]. In a prospective study of 57 patients with bladder cancers, Apolo et al. reported a sensitivity of 81% and specificity of 94% using a maximum SUV (SUVmax) of ≥ 4 for detecting locoregional lymph node involvement [66]. Vind-Kezunovic et al. selected different SUVmax cutoff values of 2, 3, 4 and 5 to evaluate the diagnostic ability of FDG PET-CT for detecting lymph node involvement based on SUVmax. Their results showed higher specificity at higher values of SUVmax (specificity of 91.1% and 97.7% for SUVmax ≥ 4 and SUVmax ≥ 5, respectively) [67].

Hybrid PET-MRI was recently introduced to incorporate the functional information of both MRI and PET with high soft tissue contrast anatomic information of MRI [68]. In a study of 134 patients with pelvic malignancies, Catalano et al. reported that FDG PET-MRI was more accurate than FDG PET-CT in locoregional lymph node staging [69]. Rosenkrantz et al. prospectively analyzed 30 FDG PET-MRI examinations using diuresis protocol in 22 bladder cancer patients. The imaging protocol for this study consisted of injecting FDG one hour prior to imaging followed by forced diuresis. Diuresis was attained by oral hydration and intravenous furosemide. PET-MRI showed higher accuracy (95%) in detecting lymph node involvement as compared to MRI (76%) alone. They suggested a clinically practical role for PET-MRI in the evaluation of patients with bladder cancer especially in clarifying findings that were equivocal on MRI alone [55].

In a recent study of 15 patients with bladder cancer by Salminen et al., the accuracy of PET-MRI in detecting muscle-invasive disease was 73%; the sensitivity, specificity and accuracy for lymph node involvement was 50%, 67% and 60% respectively. Although sensitivity for lymph node involvement was low, the overall accuracy of PET-MRI suggests a potential role for local staging of bladder cancer [70]. Sensitivity, specificity and accuracy of PET-CT and PET-MRI with different techniques and radiotracers for detecting lymph node involvement are shown in Table 6.

Table 6 PET-CT and PET-MRI characteristics for detecting lymph node involvement in patients with bladder cancer
Full size table

Recent advances and future directions

Standardize reporting

In this article, we presented different perspectives on multiple modalities for the local staging of bladder cancer. Amongst the currently available options to the radiologist, mpMRI has the potential to decrease the inter-observer discrepancy, largely though the depiction of the detrusor muscle and hence whether it is invaded with tumor. In addition to superior soft tissue characterization, the MRI derived functional imaging data that makes this modality to emerge as the locoregional staging imaging method of choice in bladder cancer (Fig. 7). The success of other imaging-based standardized reporting schemes, such as Breast Imaging Reporting and Data System (BI-RADS) and Prostate Imaging Reporting and Data System (PI-RADS) has led to efforts to standardize the MRI reporting in bladder cancer. The fruit of one of these efforts was the publication of the first version of Vesical Imaging-Reporting and Data System (VI-RADS) in 2018 [76]. The authors used the currently available anatomical and function MRI-derived evidence to derive a 5-scale certainty score from highly unlikely (VI-RADS1) to highly likely tumor invasion of muscle and tissues beyond the bladder (VI-RADS5). The four components of the imaging sequences used to establish a VI-RADS score include anatomical imaging with T2WI, and functional imaging using DCE imaging, DWI, and ADC. In this system, priority is given to the T2WI to determine the extent of muscle involvement; the functional sequences are used to further stratify the mass into VI-RADS categories. This method has been evaluated by Barchetti et al. [77] in 75 patients with bladder cancer. They demonstrated a good interobserver agreement (k = 0.731) for accurately categorizing the mass as MIBC versus non-MIBC with AUC of 0.87 and 0.92 between two observers. In a second study using VI-RADS scheme, Wang et al. [78] retrospectively evaluated 340 patients who underwent mpMRI. The area under the ROC curve for muscle invasion was 94% with sensitivity and specificity of 87.1 and 96.5, respectively. Finally, another recent report by Ueno et al. [79] investigated the role of VI-RADS reporting scheme in 74 patients with mpMRI who underwent TURBT. The diagnostic performance of VI-RADS in the differentiation of MIBC vs. non-MIBC was excellent with pooled AUC of 90%. In this study the interclass correlation coefficients (ICC) for interobserver agreements between five radiologists was excellent (ICC: 0.85). Given the excellent diagnostic performance and interobserver agreements of VI-RADS in these preliminary studies, VI-RADS scoring system may eventually become standard way to report pretreatment MRI exams in patients with bladder cancer.

Texture Analysis

There are thousands of pixels in each data set that are presented to the radiologist for interpretation. However, the human eye is capable of recognizing a limited number of shades of gray and associations between adjacent pixels. Texture analysis is a quantitative method for understanding the detailed distributions of pixels or heterogeneity in a given image. Quantitative CT-derived texture analysis has been shown to be associated with tumor grade and survival in patients with colorectal cancer [80], non-small cell lung cancer [81] and esophageal cancer [82] among others. However, given the limitations of CT for tissue characterization in bladder cancer, efforts have been focused on texture analysis of bladder masses using MRI. Recently, Lim et al. [83] investigated the relationship between texture analysis features of the bladder mass and perivesical fat, using T2WI and ADC, in patients who underwent TURBT for staging. Entropy of the mass and extravesical fat were higher for tumors that were stage ≥ T3 compared to tumors that were stage ≤ T2. Similarly, entropy was higher in the mass and extravesical fat in ≥ T2 tumors compared to T1 tumors. Others have shown a similar potential of CT-derived texture analysis in differentiating low-grade versus high-grade urothelial carcinomas [84]. Given the promise of texture analysis in other fields of cancer imaging, one can expect use of higher-order texture analysis for characterization, staging and prognostication of bladder cancer in the future.

Conclusion

Reliable preoperative staging is a critical step in establishing optimal management and treatment decisions in patients with bladder cancer. Extensive efforts are being made to improve the accuracy of both the cross-sectional and molecular imaging modalities in local staging of the bladder cancer. Based on the current evidence, MRI is the modality of choice for locoregional staging of bladder cancer with a combination of anatomic and functional pulse sequences. Despite the advantages of the MRI in the locoregional staging of the bladder cancer, CT urography remains the modality of choice for initial evaluation of patients who present with hematuria. This is mainly due to superior performance of CT in evaluation of the upper urinary tract, in addition to ease of access and convenience for the patients.