Multidetector computed tomography urography for diagnosing upper urinary tract urothelial tumour

Authors


Nigel Cowan, Department of Radiology, The Churchill Hospital, Oxford OX3 7LJ, UK.
e-mail: nigel.cowan@nds.ox.ac.uk

Abstract

OBJECTIVE

To evaluate multidetector computed tomography urography (MDCTU) for diagnosing upper urinary tract (UUT) urothelial tumour by comparison with retrograde ureteropyelography (RUP).

PATIENTS AND METHODS

MDCTU and RUP were used in a selected series of adult patients presenting with haematuria. Entry criteria were based on findings on intravenous urography and were chosen to ensure a high prevalence of UUT urothelial tumour to allow a valid retrospective comparison of the diagnostic techniques. MDCTU and RUP studies were scored for the presence and absence of UUT urothelial tumour by two radiologists, retrospectively and independently, and while unaware of the demographic and clinical information. The reference standards were the histopathology and clinical follow-up.

RESULTS

MDCTU and RUP were used in 106 patients over a 24-month period. RUP was attempted in 151 of 212 UUTs; the corresponding MDCTU for each UUT was reviewed. MDCTU was a true-positive (TP) for urothelial tumour in 31, true-negative (TN) in 111, false-positive (FP) in eight and false-negative (FN) in one UUT, giving a sensitivity of 0.97, a specificity of 0.93, a positive predictive value (PPV) of 0.79 and a negative PV (NPV) of 0.99. RUP was technically successful and diagnostic in 96% of the UUTs (143/151). For diagnosing urothelial tumour, RUP was TP in 26, TN in 112, FP in four and FN in one UUT, giving a sensitivity of 0.97, specificity of 0.93, a PPV of 0.79 and NPV of 0.99.

CONCLUSION

This study validates quantitatively the use of MDCTU for diagnosing UUT urothelial tumour.

Abbreviations
MDCT(U)

multidetector CT (urography)

UUT

upper urinary tract

RUP

retrograde ureteropyelography

EAU

European Association of Urology

P(N)PV

positive (negative) predictive value

MPR

multiplanar reformatted.

INTRODUCTION

Multidetector CT (MDCT) is an established imaging method for detecting and characterizing renal masses, and identifying urinary tract calculi [1,2]. MDCT urography (MDCTU) is a more recent technique developed primarily for investigating urothelial lesions, renal masses and urinary tract stones [3]. MDCTU can also detect significant extra-genitourinary pathology [4]. Multiplanar image analysis is possible with MDCTU, as the thin slices make isotropic or near-isotropic imaging feasible. The concept of MDCTU is attractive as both the renal parenchyma and urothelium can be evaluated with one relatively noninvasive comprehensive examination. Initial experience indicates that MDCTU is a promising method for investigating benign and malignant lesions in the upper urinary tract (UUT) [3].

The most recent AUA guidelines (2001) on the evaluation of asymptomatic haematuria in adults [5], state that although high detection rates of TCC were reported for contrast-enhanced CT images [6,7] the data were considered anecdotal because the studies offered no statistical analysis. The AUA guidelines also indicate that retrograde ureteropyelography (RUP) is commonly considered by many to be the best imaging approach for detecting TCC, but again this opinion is not based on evidence [5].

The latest European Association of Urology (EAU) guidelines on the diagnosis and treatment of UUT TCC (2004) [8] state that IVU is still the first choice of examination for investigating haematuria and that RUP might be useful in cases where IVU is equivocal, with a sensitivity of >75%[9,10]. The EAU guidelines suggest that CT might be useful in the diagnosis and staging of renal parenchymal tumours, but suggest that with CT it is difficult to accurately diagnose small-volume tumours of the renal pelvis and ureter [11,12]. The EAU guidelines recognize that CT can determine the local extent of the primary tumour, invasion in renal parenchyma, and presence of lymph nodes or liver metastasis better than any other imaging technique [11,12], and therefore only advocate CT when an invasive tumour must be excluded. However, the EAU guidelines refer to single-slice CTU and should be updated to be truly relevant to current clinical practice which uses MDCT.

Thus the purpose of the present study was to evaluate quantitatively the use of MDCTU for the diagnosis of UUT urothelial tumour in patients with haematuria, by comparison with RUP, using histopathology and clinical follow-up as the reference standard.

PATIENTS AND METHODS

MDCTU and RUP were used to diagnose and stage UUT urothelial tumours in a selected series of patients presenting with haematuria. Initial investigations included IVU and flexible cystoscopy in all. Patients with equivocal or positive IVU findings, and those with persistent haematuria, negative IVU and flexible cystoscopy, were investigated for UUT urothelial tumour with MDCTU and RUP. The study was conducted over a 24-month period.

MDCTU and RUP studies of the UUT were reviewed retrospectively and independently by two radiologists with no knowledge of the demographic details or clinical information. A decision relating to the presence or absence of UUT urothelial tumour was reached. The results were analysed using 2 × 2 tables and the sensitivity, specificity, positive and negative predictive values (PPV, NPV) were calculated for MDCTU and RUP for diagnosing UUT urothelial tumour. The reference standards were histopathology (obtained by biopsy or from resected specimens) and the clinical follow-up, by reviewing the medical and pathological records over a 3–5-year period.

CT was conducted on a multidetector scanner (GE Lightspeed QX/i, GE Medical Systems, Milwaukee, WI, USA) using eight slices. All patients were given 500–750 mL of water to drink in the 20 min before CT; no oral contrast medium was administered. The patient was placed supine on the CT table. Phase 1 was unenhanced from the top of the kidneys to 2 cm below the symphysis pubis on expiration, using 8.0 × 2.5 mm collimation. A double-bolus of i.v. contrast medium was used, which was a variant of the protocol described by Chow and Sommer [13]. Phase 2 was a combined nephrographic-pyelographic phase. A bolus of 100 mL of non-ionic contrast medium (Iopamidol 300) was given i.v. by hand at 2–3 mL/s and the patient was exercised in the CT room by walking and then touching their toes, and finally by rolling on the table 720° in both clockwise and anticlockwise directions, to thoroughly mix the contrast medium with urine in the UUT and bladder. At 10 min after the first injection, a second bolus of 50 mL of the same contrast medium was given via a pump at 3 mL/s, and the abdomen and pelvis scanned in expiration 100 s after the second bolus, producing a combined nephro-pyelographic phase; the scanning parameters are listed in Table 1.

Table 1.  The scanning parameters for MDCTU
VariableProtocol ID (8), MDCTU
Before contrast+Contrast*2nd R3rd R
  1. SC, slice collimation; BC, beam collimation; TS, table feed; SFOV, scanned field of view; DFOV, display field of view; SW, section width; RI, reconstruction interval; n, detector configuration. *First reconstruction for early review; †2nd reconstruction for axial review; ‡3rd reconstruction for reporting and three-dimensional display.

StartTop of kidneyTop of liverTop of liverTop of liver or kidney
EndBottom of symphysisBottom of symphysisBottom of symphysisBottom of symphysis
n  1.25 × 8  1.25 × 8  1.25 × 8  1.25 × 8
SC, mm  1.25  1.25  1.25  1.25
BC, mm 10 10 10 10
TS, mm/rot 16.75 16.75 16.75 16.75
Pitch  1.675  1.675  1.675  1.675
SFOV, mm500500500500
Matrix512512512512
DFOV, mm320320320320
SW, mm  2.5  5  2.5  1.25
RI, mm  1.25  5  1.25  0.63
kV120120120120
Auto mA190–220100–300100–300100–300

If the UUT opacification was incomplete when the first axial reconstruction was reviewed by the attending radiologist, a second delayed excretory phase scan was taken with the patient prone; this position was selected to encourage complete UUT opacification and further mixing of contrast medium with urine. The unenhanced scans were reconstructed axially using 2.5-mm sections and 1.25-mm increments. The nephro-pyelographic phase images were reconstructed axially, initially using 5-mm sections at 5-mm intervals, and then using 2.5-mm sections and 1.25-mm increments, and at 1.25-mm sections and 0.625 mm increments in selected cases where multiplanar reformatted images were viewed to achieve isotropic voxels.

For the RUP, informed consent was obtained from all patients and antibiotic prophylaxis (ciprofloxacin 500 mg orally) given. Lidocaine gel was used as a local anaesthetic and lubricant. Sedoanalgesia (diazemuls, 2.5–10 mg i.v., and pethidine 50–100 mg i.v.) was administered as required. The patient’s pulse, blood pressure and oxygen saturation were continuously monitored during the procedure.

A flexible cystoscope was passed into the bladder and rotated through 180° to allow greater deviation of the end of the cystoscope and facilitate identification of the ureteric orifices. A 0.09 mm straight hydrophilic guidewire (Terumo Corporation, Tokyo, Japan) was passed into the ureteric orifice under direct vision. The guidewire was manipulated into the renal pelvis using C-arm digital fluoroscopy for guidance (Siemens Polystar, Erlangen, Germany). The flexible cystoscope was removed and a 4 F general-purpose vascular catheter (Cordis, Miami, FL, USA) placed over the wire into the renal pelvis. RUP was then performed using C-arm rotation and the table-tilting facility [14]. Low osmolar non-ionic contrast medium was used (Iopamidol 300) and diluted if appropriate with normal saline.

Examinations were reviewed and scored by two radiologists, with no clinical information or knowledge of the patients’ demographics. MDCTU, axial, multiplanar reformatted (MPR) images were constructed and reviewed at a workstation, running Voxar3D version 4.2 (Voxar, Edinburgh, UK) on ‘abdominal’ and ‘bone’ window settings. Using axial review of phase 1 and then phase 2, a diagnosis could be made in most cases. MPR analysis was used in specific cases only for clarification. The RUP images were also reviewed on a workstation; individual UUTs were reviewed retrospectively and independently by two radiologists and a consensus score obtained. MDCTU and RUP examinations were reviewed 3 months apart.

RESULTS

MDCTU and RUP were compared in 151 UUT in 106 patients (mean age 64.9 years, range 25.1–90.5; 71 men and 35 women) with haematuria (77 macroscopic and 29 microscopic) over a 24-month period. MDCTU was technically successful, providing images of diagnostic quality, in all 151 UUTs. RUP was technically successful in 96% (145/151) of the UUTs attempted.

The aim of RUP is to provide diagnostic information about the entire UUT so that an appropriate clinical management plan can be made. Tumour present in the bladder might prevent an adequate RUP study but does not necessarily indicate UUT involvement. In the present study, six (4%) attempted RUPs were technically unsuccessful; UUT urothelial tumour was present in four and absent in two. The reasons for technical failure were bladder tumour obscuring the ureteric orifice in four, blood obscuring the orifice in one and tumour obstructing the lower ureter, preventing passage of the guidewire, in one.

Of all the RUP studies, two were classified as technically successful but not diagnostic; in these two there was complete occlusion of the ureter, making a diagnosis from the RUP images alone impossible. In one of the cases TCC was responsible for the ureteric occlusion, and in the other it was due to PUJ obstruction secondary to chronic pyelonephritis. Thus, of the 151 RUP studies attempted, 143 were technically successful and of sufficient diagnostic quality to be included in the quantitative analysis.

The prevalence of UUT urothelial tumour in the patients assessed was 30.2% (32/106). Such a high prevalence is explained by the inclusion criteria for the study. The sites and frequency of UUT urothelial tumour are shown in Table 2.

Table 2.  UUT urothelial tumour site and frequency
SiteNumber
  1. There were three other UUT tumour types, i.e. two RCC and one metastases to kidney from adenocarcinoma of the lung.

TCC kidney 8
TCC renal pelvis 9
TCC ureter – multifocal 2
TCC ureter – unifocal13
Total32

Analysing the results for MDCTU compared with the reference standard of histopathology and follow-up showed that MDCTU had a sensitivity of 0.97 and a specificity of 0.93, with a PPV of 0.79 and NPV of 0.99 for the diagnosis of UUT urothelial tumour (Table 3). MDCTU was false-negative for urothelial tumour in one UUT; the lower ureter was incompletely opacified at the site of a small urothelial tumour. Diagnosis from the MDCTU was not possible from the excretory phase images. MDCTU was false-positive for urothelial tumour in eight UUT. Debris in the collecting system was misinterpreted as tumour in three. Circumferential ureteric wall thickening was mistaken for tumour in two. One of these cases was secondary to an iatrogenic injury at ureteroscopy for stone removal, and the other showed slight, circumferential wall thickening at the site of a ureteric kink.

Table 3. 
The evaluation of MDCTU and RUP for diagnosing UUT urothelial tumour
MethodTumour positiveTumour negativeTotal
  1. Technical failures and technically successful but not diagnostic studies were excluded from the analysis. For MDCTU, sensitivity 0.97, specificity 0.93, PPV 0.79, NPV 0.99; for RUP, sensitivity 0.96, specificity 0.97, PPV 0.87, NPV 0.97.

MDCTU
 positive31  8 39
 negative 1111 112
 total32 119 151
RUP
 positive26  4 30
 negative 1 112 113
 total27 116143

RCC with collecting system invasion was interpreted as TCC in one case. A vessel causing an indentation in an upper pole infundibulum was interpreted incorrectly as TCC in one other, and the final false-positive MDCTU showed a small filling defect on axial sections in the lumen of the lower ureter close to the vesico-ureteric junction. The corresponding RUP showed a tumour-free ureter with ‘fish-hooking’ as it passed over an exophytic bladder cancer in one case.

Analysing the results for RUP compared with histopathology and follow-up, with the technically inadequate studies excluded from the analysis, RUP had a sensitivity of 0.96 and a specificity of 0.97, with a PPV of 0.87 and NPV of 0.97 for the diagnosis of UUT urothelial tumour (Table 3). RUP was false-negative for one UUT urothelial tumour when there was circumferential ureteric wall thickening with no epithelial irregularity, which was detectable only on MDCTU. RUP was false-positive for UUT urothelial tumour in four cases. A vascular impression caused irregularity of the inferior margin of an upper pole infundibulum of the right kidney in one. Subsequently CT arteriography confirmed these signs were due to a serpiginous vessel from as small arteriovenous fistula. Irregularity of the epithelium in the region of the PUJ was shown by MDCTU to be secondary to an impacted calculus dislodged and not identified at RUP in one case. A solid RCC and a renal metastasis from adenocarcinoma of the lung were tumours that mimicked UUT urothelial tumour, both showing invasion of the collecting system.

DISCUSSION

In this study we evaluated quantitatively the use of MDCTU for diagnosing UUT urothelial tumour. Currently there are very few studies comparing the performance of various diagnostic imaging methods for diagnosing TCC in the UUT [15,16]. High detection rates for TCC on contrast-enhanced CT images are suggested [3,17,18], but such reviews have no statistical analysis. The AUA Best Practice Policy Recommendations of 2001 state that RUP is generally considered the best imaging method for detecting and characterizing ureteric abnormalities [5]. In the present study we evaluated MDCTU for diagnosing UUT urothelial tumour and provide a statistical analysis. MDCTU was compared with RUP because RUP is currently assumed to be the best diagnostic test for these tumours.

MDCTU and RUP have similar high sensitivities for diagnosing UUT urothelial tumour; sessile or pedunculated TCC might be detected with similar sensitivity by both MDCTU and RUP. Those urothelial tumours that show circumferential urothelial thickening with epithelial irregularity (Fig. 1) might also be diagnosed by both diagnostic techniques. However, MDCTU can detect urothelial wall thickening with no epithelial irregularity. Such cases of UUT urothelial tumour will be missed by RUP (Fig. 2).

Figure 1.

(a) Circumferential urothelial thickening of the ureter seen on an excretory phase CTU caused by TCC; (b) Coronal MPR image of an excretory phase CTU showing circumferential urothelial thickening of the ureter; (c) RUP showing narrowing and minor but definite irregularity of the epithelium of the upper ureter from TCC.

Figure 2.

(a) Axial excretory phase CTU showing circumferential wall thickening of the right ureter caused by TCC. (b) Excretory phase CTU; coronal MPR image showing circumferential wall thickening of the right ureter from TCC; (c) RUP showing a smooth epithelium, masking the presence of urothelial tumour.

The one case of false-negative MDCTU for UUT urothelial tumour was due to incomplete ureteric opacification by contrast medium. The diagnosis of UUT urothelial tumour depends on the difference in density between the tumour and surrounding contrast medium. In ureteric segments in which there is incomplete opacification the difference in density between the tumour and ureteric wall might be so small as to render small tumours undetectable (Fig. 3). Complete and homogenous opacification of the collecting system and ureter with contrast medium is therefore desirable for optimum sensitivity of MDCTU for diagnosing UUT urothelial tumour.

Figure 3.

(a) Excretory phase CTU showing no opacification of the right ureter, with no dilatation; (b) RUP showing a lower ureteric filling defect confirmed as TCC on ureteroscopic biopsy.

The most common technical error of MDCTU is incomplete ureteric opacification, most frequently of the distal ureter [19]. Various manoeuvres can promote complete opacification of the ureter, including oral hydration before scanning [20–22], i.v. frusemide (5–10 mg), exercising the patient immediately before scanning the excretory phase [23], test scanning at defined levels before the excretory phase [24], and undertaking further series [18,25].

There is the theoretical risk of missing tumour at CTU due to incomplete mixing of contrast medium with urine within the renal pelvis, ureters and bladder. Rolling the patient on the CT table and exercising the patient just before acquiring the excretory-phase series encourages homogenization of contrast medium and urine, aiming to increasing the sensitivity of MDCTU for the diagnosis of UUT urothelial tumour.

The MDCTU false-positive results for UUT urothelial tumour fall into three groups; those with urothelial abnormalities, those with luminal abnormalities and those with extra-urothelial abnormalities, simulating UUT urothelial tumour. For the first, in the present study an iatrogenic ureteric injury at ureteroscopy led to circumferential urothelial thickening and was misdiagnosed as UUT urothelial tumour (Fig. 4). With no history, denied to the reviewers under the conditions of the study, this might be suspected. In the other case, mild circumferential urothelial thickening was identified at the site of a ureteric kink. A routine MPR review would assist in differentiating UUT tumour thickening from artefactual wall thickening. Other causes of urothelial wall thickening include irritation by calculi or stent, fibroepithelial polyp or rare tumours such as the nephrogenic adenoma.

Figure 4.

(a) Axial excretory phase CTU; circumferential wall thickening of the left lower ureter was interpreted as TCC. In this study, a history of a difficult ureteroscopic stone removal was not available when the image was interpreted; (b) RUP showing a raised urothelial flap and a short lower ureteric stricture after ureteroscopy for stone removal, but no evidence of TCC.

For luminal abnormalities, in four UUT debris was mistaken for UUT urothelial tumour (Fig. 5). Lack of enhancement would help to differentiate debris from tumour, as might repositioning and re-scanning with the patient prone, and at the specific site of interest, to see if the debris shifted with repositioning. Intraluminal clot could be differentiated from tumour using these techniques.

Figure 5.

(a) Axial excretory phase CTU showing hydronephrosis of the right kidney and many small filling defects caused by debris, misinterpreted as multifocal TCC; (b) RUP showing tight PUJ obstruction and hydronephrosis of the right kidney. The filling defects seen on CTU within the collecting system were not identified at RUP.

For the last group, vascular indentation on an upper pole infundibulum is difficult to distinguish from tumour (Fig. 6). CT arteriography or RUP would help to clarify this situation. Finally, other tumour types sometimes mimic TCC, e.g. RCC when invading the collecting system (Fig. 7). The only method of determining the exact cell type under these conditions is by biopsy and histopathology review.

Figure 6.

(a) Axial excretory phase CTU; the small impression on the upper pole infundibulum (arrow) was recorded as urothelial tumour; (b) RUP showing the upper pole infundibulum with smooth indentation (arrow) characteristic of a vascular impression.

Figure 7.

(a) Axial excretory phase CTU showing a soft-tissue density mass indenting the infundibulum. The patient had had a right nephrectomy for RCC. This was reported as a TCC without the benefit of the clinical history; (b) RUP showing a filling defect with a smooth margin indenting the superior part of the renal pelvis. The lesion was reported as a TCC.

In principle MDCTU has many advantages over other imaging methods for investigating patients with haematuria who are at risk of urological malignancy. MDCTU might be used as a single noninvasive test for examining the entire urinary tract. It provides information on the presence of stones, urothelial tumours, renal tumours and extra-genitourinary pathology [4]. Recent work suggests that it has a role for diagnosing bladder tumours [26]. Compared with RUP, MDCTU, as it is noninvasive, offers less chance of physical trauma, e.g. rupture of the collecting system, introduction of infection or irritation to the urinary tract.

The principal disadvantage of MDCTU is the increase in radiation dose to the patient when compared with IVU or RUP [27,28]. For patients with possible malignancy the increased radiation risk of MDCTU is justified because it has equivalent sensitivity and specificity for the diagnosis of UUT urothelial tumour when compared with RUP, but offers the additional benefits of being noninvasive, quicker, less labour-intensive, cheaper, with fewer complications, and allows simultaneous diagnosis and staging of UUT tumours.

For young patients in whom malignancy is unlikely and for those with benign disease, MDCTU can be used as a last resort test only if the other imaging tests are negative, and with continuing symptoms or equivocal findings [25].

The most effective method to reduce the radiation dose is to use fewer acquisitions or series. If MDCTU could be limited to one series through the abdomen and pelvis, or unenhanced through the kidneys followed by one series after i.v. contrast medium including the abdomen and pelvis, then the effective dose of MDCTU would fall within the range of conventional urography [28], making MDCTU an even more attractive method.

The present study provides a quantitative evaluation of MDCTU for detecting UUT urothelial tumour in adults presenting with haematuria. The results are significant because the increased sensitivity and similar specificity of MDCTU for detecting UUT urothelial tumour compared with RUP means that CTU is now used before RUP in our institution. RUP is no longer used to clarify equivocal IVU or ultrasonography results, but instead we use MDCTU, a change in practice brought about by the present work. In those centres with MDCTU, it is often used as a problem-solving technique when the other imaging methods provide equivocal or normal results in the presence of continuing haematuria [25]. The present study adds to the evidence that MDCTU could be used as a first-line investigation in those patients where the risk of disease outweighs the risk of radiation exposure. The most appropriate patients are those at high risk of urological cancer. If MDCTU is used as a first-line test, ultrasonography and IVU can be avoided and RUP used only if the MDCTU findings are equivocal or not diagnostic, e.g. in the unusual case of incomplete ureteric opacification. The overall result of making this readjustment is an acceleration of the diagnostic imaging pathway for diagnosing UUT urothelial tumour.

The method of RUP used was unusual, as it was done with the patient under sedoanalgesia in the radiology department, using high-quality digital C-arm fluoroscopy [14]. The RUP images are of higher quality than those obtained using a mobile unit. As this method is not widely used, it makes the present study difficult to replicate. RUP was followed by ureteroscopy and biopsy when RUP was positive for urothelial tumour. A few reports suggest that ureteroscopy and biopsy are more sensitive than RUP for diagnosing urothelial lesions [10]. It is assumed that a 3–5-year clinical follow-up, as used here, will be sufficient to allow the diagnosis of all UUT urothelial tumours in the population.

In conclusion, MDCTU and RUP have similar diagnostic sensitivity and specificity for diagnosing UUT urothelial tumour, which validates quantitatively the use of MDCTU. MDCTU should therefore be used before RUP, as it is a single noninvasive comprehensive test that allows simultaneous diagnosis and/or staging. RUP should be restricted to patients with a non-diagnostic MDCTU (usually due to incomplete ureteric opacification) or with impaired renal function, and hence a predisposition to contrast-induced nephropathy [29].

If MDCTU potentially replaces ultrasonography, IVU and RUP for investigating haematuria in specific groups of patients in whom the increased radiation dose is justified, the imaging pathway for diagnosing UUT urothelial tumour might be accelerated by reducing the number of diagnostic episodes with no reduction in diagnostic accuracy [27].

ACKNOWLEDGEMENTS

We thank Miss Sarah-Jane Holt and Miss Heidi Musk for their work in collating the images for review.

CONFLICT OF INTEREST

None declared.

Ancillary