The aim of this study was to determine the diagnostic accuracy, especially the false negative rate (FNR), of CTU with CMP in exclusion of UBC and whether some patients could omit cystoscopy.
Patients
All patients evaluated for macroscopic hematuria according to SCP in the NU Hospital Group, Uddevalla, Sweden, between 1st November 2016 and 31st December 2019 were retrospectively included. The criteria for SCP, and therefore the inclusion criteria of the study, were macroscopic hematuria and ≥ 40 years age. The age criterion was changed to ≥ 50 years in 2018. Patients were excluded from the study if they did not undergo CTU, or if the CTU was not done with CMP or UP (Fig. 1).
The study group was divided into the UBC group, which consisted of all patients diagnosed with UBC during the study period, while a random sample of patients who had macroscopic hematuria but without detected UBC, matched by year of diagnosis, constituted the control group. From this study group the results were generalized to the study cohort.
Medical records were retrospectively reviewed to retrieve tumor-specific data for the UBC group (e.g., tumor number and tumor size which were estimated during TURBT using the loop of the resectoscope (7 mm as a reference) and to ensure that patients in the control group were not diagnosed with UBC until 31st March 2021. All UBC patients underwent TURBT and had a histopathologically verified UBC diagnosis.
Imaging technology
Preparation before examination included drinking of 1000 mL of water and not voiding approximately 90 min prior to examination. CTU was performed with the patient in supine position, using a 64-detector scanner (General Electric, Boston, USA or Siemens Medical solutions, Forchheim, Germany). A four-phase protocol including UP of the abdomen and pelvis (70–480 mA) was initiated. This was followed by a CMP of the abdomen and pelvis (120–560 mA) at bolus tracking + 20 s after intravenously administration of iodinated contrast medium (Iohexol 350 mg/mL; 400 mg I/kg, 20 g I/kg/sec, Omnipaque; GE Healthcare, Waukesha, WI, USA).
The following phase was a nephrographic phase (NP) of the kidneys (120–560 mA) at bolus tracking + 40 s. After a short mobilization of the patient, an excretory phase (EP) of the abdomen and pelvis (70–480 mA) > 7.5 min after contrast medium administration was obtained. All four phases were reduced with 40% mAs with GE or made with a quality reference mAs Siemens. In other words, the UP and EP could be considered as “low-dose” phases and the CMP as a “normal-dose” phase concerning radiation. Collimation of 0.6 mm, pitch of 1.4 and 120 kVp were applied in all phases. No diuretic drugs were given.
Imaging analysis
All included CTU scans were copied to a separate study database, pseudonymized and purged of all annotations. Two specialists in radiology (CH and DV) with seven and eighteen years of experience of interpretation of CTU who were blinded to all clinical data and previous CTU results, reviewed the pseudonymized CTUs independently. The number of tumors and the largest size of the largest tumor were recorded. In addition, inadequacy of bladder filling volume, existence of bladder stones, indwelling catheter, existence of thickened bladder wall and other image distortions that made the re-evaluation difficult, such as artifacts due to hip prostheses, were also noted separately.
The reviewers categorized the CTUs as either positive (showing UBC) or negative (not showing UBC). The negative CTUs were then further categorized as having good image quality (good/very good bladder filling, no non-specific bladder wall thickening, no indwelling catheter, no bladder stones, and no significant image distortions) or having limited image quality (having one or more of the above). The CTUs were thereby categorized as positive (POS), indeterminate (not showing UBC but with limited quality, IND), or negative (not showing UBC with good quality, NEG). After the first round of review, a consensus between the two radiologists was reached in a joint review in the cases where the interpretations differed. All statistical analyses were based on this consensus interpretation if not otherwise indicated. After unblinding, one further round of review of the cases with false negative CTU in the group (NEG) was performed by a senior radiologist (HL), who was not blinded to the results of the other radiologists interpretations to try to identify possible systematic errors in interpretation.
Statistical analysis
All available UBC patients were included in the study. To be able to demonstrate an expected FNR of around 10% and a false positive rate (FPR) < 50% the required sample size was calculated to approximately 200 UBC patients and 100 controls [14]. The control group was increased by approximately 10% to account for exclusions and were randomly selected from the patients without UBC, matched by year of diagnosis, for a total of 113 patients. This sampling of the controls was done to keep the number of re-evaluations manageable and focused on the FNR, while still being able to estimate other accuracy measures and omittable cystoscopies. This procedure allowed us to generalize the results from the study group to the study cohort with highly valid results.
Descriptive statistics were used for patient characteristics (age at CTU and sex) and tumor characteristics (clinical tumor stage, tumor size and number of tumors). Continuous data were presented as mean with standard deviation (SD). Inter-rater variability of the interpretations of the CTUs was analyzed with Cohen’s kappa with > 0.80 interpreted as good agreement. Receiver operating characteristic analysis was done and the area under the curve (AUC) was calculated.
The FNR and FPR were calculated with 95%-confidence intervals (CI) according to the score method with continuity correction described by Wilson [15]. Positive predictive value (PPV) and negative predictive value (NPV) were calculated with 95% CI similarly, and were based on the entire cohort of patients with macroscopic hematuria, i.e., results of the control group were multiplied to represent the entire cohort without UBC, assuming similar results in randomly selected controls and in those patients without UBC who were not evaluated in the present study. Statistical analysis was performed using SPSS version 27 (IBM Corp., Armonk, NY, USA).
Omittable cystoscopies was defined as the POS CTUs, where the patient could go directly to a TURBT, and the NEG CTUs. This was calculated as a proportion of the entire cohort of patients with macroscopic hematuria, i.e., the randomly selected control group was generalized to reflect the entire group of patients without UBC, and the patients that had not undergone a CTU according to the protocol were counted as requiring cystoscopy. The number of cystoscopies needed to detect one case of UBC, and one case of high-risk UBC, respectively, among the NEG CTUs were calculated.