BJU Int 2022; 130: 754–763 doi:10.1111/bju.15677 Original Article Nomograms including the UBC® Rapid test to detect primary bladder cancer based on a multicentre dataset Christina J. Meisl1 , Pierre I. Karakiewicz2, Roland Einarsson3, Stefan Koch4,5, Steffen Hallmann6, Sarah Weiß6, 7 Tammer Hemdan , Per-Uno Malmstro € m7 , Johan Styrke8 , Amir Sherif8, Mudhar N. Hasan9, Renate Pichler10 , 11  € rg Hennenlotter12, Arnulf Stenzl12  Ritter13, Gennadi Tulchiner , Joan Palou , Oscar Rodrıguez Faba 10  11 , Jo , Rene Gu€ nter Niegisch14, Camilla M. Grunewald14, Thorsten Schlomm1, Frank Friedersdorff1,15, Dimitri Barski16, Thomas Otto16,17, Andreas Go € ssl18, Christian Arndt18, Kesavan Esuvaranathan19, Nisha R. Kesavan19, Zang Zhijiang19, Mario W. Kramer , Martin J. P. Hennig20 and Thorsten H. Ecke1,6 20 1 Department of Urology, Charite  – Universitatsmedizin € Berlin, Corporate Member of Freie Universitat € Berlin, Humbold- € zu Berlin, and Berlin Institute of Health, Berlin, Germany, 2Department of Surgery, University of Montreal, Universitat Montreal, QC, Canada, 3IDL Biotech, Bromma, Sweden, 4Department of Pathology, HELIOS Hospital Bad Saarow, Bad Saarow, 5Brandenburg Medical School Theodor Fontane, Neuruppin, 6Department of Urology, HELIOS Hospital Bad Saarow, Bad Saarow, Germany, 7Department of Surgical Sciences, Uppsala University, Uppsala, 8Department of Surgical and Perioperative Sciences, Urology and Andrology, Ume a University, Umea , 9Department of Urology, 10 Danderyd Hospital, Danderyd, Sweden, Department of Urology, Medical University Innsbruck, Innsbruck, Austria, 11 Department of Urology, Fundacio  Puigvert, Autonomous University of Barcelona, Barcelona, Spain, 12Department of € bingen, Tu Urology, University Hospital Tu € bingen, 13Department of Urology, Diakonie-Klinikum Stuttgart, Stuttgart, 14 Department of Urology, Faculty of Medicine at Heinrich Heine University Du € sseldorf, Du€ sseldorf, 15Department of Urology, Evangelisches Krankenhaus Konigin€ Elisabeth Herzberge, Berlin, 16Department of Urology, Rheinland Klinikum Neuss, Neuss, 17Medical Faculty, University of Duisburg-Essen, Essen, 18Department of Urology, HELIOS Hospital Krefeld, Krefeld, Germany, 19Department of Urology, National University Hospital, Singapore City, Singapore, and 20Department of Urology, University Hospital Schleswig-Holstein Campus Lu € beck, Lu € beck, Germany Objectives To evaluate the clinical utility of the urinary bladder cancer antigen test UBCâ Rapid for the diagnosis of bladder cancer (BC) and to develop and validate nomograms to identify patients at high risk of primary BC. Patients and Methods Data from 1787 patients from 13 participating centres, who were tested between 2012 and 2020, including 763 patients with BC, were analysed. Urine samples were analysed with the UBCâ Rapid test. The nomograms were developed using data from 320 patients and externally validated using data from 274 patients. The diagnostic accuracy of the UBCâ Rapid test was evaluated using receiver-operating characteristic curve analysis. Brier scores and calibration curves were chosen for the validation. Biopsy-proven BC was predicted using multivariate logistic regression. Results The sensitivity, specificity, and area under the curve for the UBCâ Rapid test were 46.4%, 75.5% and 0.61 (95% confidence interval [CI] 0.58–0.64) for low-grade (LG) BC, and 70.5%, 75.5% and 0.73 (95% CI 0.70–0.76) for high-grade (HG) BC, respectively. Age, UBCâ Rapid test results, smoking status and haematuria were identified as independent predictors of primary BC. After external validation, nomograms based on these predictors resulted in areas under the curve of 0.79 (95% CI 0.72–0.87) and 0.95 (95% CI: 0.92–0.98) for predicting LG-BC and HG-BC, respectively, showing excellent calibration associated with a higher net benefit than the UBCâ Rapid test alone for low and medium risk levels in decision curve analysis. The R Shiny app allows the results to be explored interactively and can be accessed at www.blucab-index. net. Conclusion The UBCâ Rapid test alone has limited clinical utility for predicting the presence of BC. However, its combined use with BC risk factors including age, smoking status and haematuria provides a fast, highly accurate and non-invasive tool for screening patients for primary LG-BC and especially primary HG-BC. Ó 2022 The Authors. wileyonlinelibrary.com BJU International published by John Wiley & Sons Ltd on behalf of BJU International.. www.bjui.org This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes. 1464410x, 2022, 6, Downloaded from https://bjui-journals.onlinelibrary.wiley.com/doi/10.1111/bju.15677 by Readcube (Labtiva Inc.), Wiley Online Library on [30/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Nomograms including the UBCâ Rapid test Keywords bladder cancer, diagnosis, logistic model, nomogram, UBCâ Rapid test, urothelial cancer, #uroonc, #BladderCancer, #blcsm rate of false-positive cases, their diagnostic accuracy is Introduction controversial and their use is not recommended in the Urinary bladder cancer (BC) is the tenth most prevalent current guidelines [9–11]. The UBCâ Rapid test (IDL cancer, affecting mainly the elderly and men [1]. The global Biotech, Bromma, Sweden) is a quantitative incidence and mortality rates for BC are fourfold higher in immunochromatographic POC assay that measures soluble men than in women (9.6 and 3.2 per 100 000 in men vs. 2.4 fragments of cytokeratin-8 and -18 in urine, and achieves a and 0.9 per 100 000 in women, respectively) and it shows a sensitivity of 53%–71% and a specificity of 61%–96% [12–19]. peak incidence at the age of 70–79 years [1,2]. Tobacco The UBCâ Rapid test shows particularly good results in the smoking is the most relevant risk factor for BC, with detection of high-grade (HG) BC and CIS [12–19]. Despite summary odds ratios of 3.14 for current smokers and 1.83 for efforts to quantify the influence of known risk factors on BC former smokers [3]. Even 20 years after quitting, ex-smokers and to assess the quality of diagnostic tools, their combined have a 50% higher risk of developing BC than non-smokers contribution, in which confounding effects are eliminated, is [3]. more difficult to measure. This study investigated the diagnostic accuracy of the UBCâ Rapid test for BC and The suspicion of BC often arises after the detection of proposes models for the prediction of primary BC based on microscopic or macroscopic (gross) haematuria during a the UBCâ Rapid test in combination with established patient’s examination [4]. Although the need for examination predictors of BC. The findings were validated in an of all patients with macroscopic haematuria for the presence independent external validation cohort. of BC is indisputable, this evaluation in patients with microscopic haematuria is debatable [5]. A multicentre study including patients with newly diagnosed BC found that, Patients and Methods although patients with macroscopic haematuria presented Patients with a higher stage of disease than patients with microscopic This retrospective study included 1877 patients recruited from haematuria, the groups had a similar tumour grade 13 study centres. Of these, 90 were excluded because of lack distribution [4]. BC is typically diagnosed by cystoscopy and of data or unclear tumour status. The patient population used treated by complete transurethral resection of all visible to determine the overall diagnostic accuracy of the UBCâ lesions and pathological examination. The current guidelines Rapid test comprised tumour patients with confirmed newly define cystoscopy and cytology as the standard procedures for diagnosed BC and patients with recurrent BC. The control the detection and surveillance of BC [6]. Cystoscopy is a group included patients with no history of BC (healthy minimally traumatic and reliable procedure that can detect controls) and patients with a history of BC but no evidence most cases of BC; however, it is costly and invasive, and is of disease (NED). For the development of the risk assessment associated with complications such as UTI [7]. Furthermore, nomograms for primary BC, the patient population was cystoscopy has limited sensitivity for flat lesions such as narrowed down to patients with primary BC and the control carcinoma in situ (CIS), and this invasive procedure may lead group to healthy controls. Patient data were collected to patient discomfort [8]. Urine cytology is, in combination independently at each study centre between 2012 and 2020 with cystoscopy, the most commonly used method for the and pooled in October 2020. early detection of BC, and has a high specificity of 83%–100% [9]. However, it has a low sensitivity, especially for low-grade The protocol included cystoscopy as the ‘gold standard’ for (LG) BC, where it is in the range of 4%–31% compared with detecting BC in patients with tumours and patients with 20%–53% for all patients [9]. NED. In cases of cystoscopy-verified tumours, transurethral resection of the bladder or biopsies were performed; samples An accurate urinary marker for BC could be valuable in were assessed for tumour grade according to the 2004 identifying high-risk patients and reducing the number of WHO classification [20] and for clinical stage according to control cystoscopies. The broad spectrum of point-of-care the 2009 TNM classification [21]. Exclusion criteria were (POC) urinary analysis tests currently available facilitates the any kind of mechanical manipulation (surgery, biopsy, rapid, non-invasive and cost-efficient determination of instillation, catheterization) within 10 days before urine urinary markers. However, because these tests have a high Ó 2022 The Authors. BJU International published by John Wiley & Sons Ltd on behalf of BJU International. 755 1464410x, 2022, 6, Downloaded from https://bjui-journals.onlinelibrary.wiley.com/doi/10.1111/bju.15677 by Readcube (Labtiva Inc.), Wiley Online Library on [30/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Meisl et al. sampling, UTIs, kidney and bladder stones, neobladder, and curves, Brier scores, decision curve analysis and cut-off pregnancy. analysis. Urine samples were analysed with the POC UBCâ Rapid test A significance level of a = 0.05 (two-tailed) was applied to all (IDL Biotech). The results were quantified by the Concileâ Ω P values. Statistical analysis was performed using R (version 100 POC reader (Concile, Freiburg/Breisgau, Germany). The 4.0.2 R Project for Statistical Computing) within Rstudio Inc. tests were performed according to the manufacturer’s (Boston, MA, USA). protocols. In several study centres the UBCâ Rapid test was performed several times simultaneously for each patient. This Results was due to varying study protocols. The results from these Regarding the clinical characteristics of patients, the cohort centres were then averaged. Table 1 provides an overview reflected the population of patients suspected of having BC, of patient numbers and study approvals by the local with a high proportion of men (76%) and senior patients ethics committees of the respective study centres. Written overall (Table 2). The optimal threshold for the quantitative consent was obtained from each patient prior to study UBCâ Rapid test, according to the Youden index, was a cut- enrolment. off ≥9.88 µg/L. The results of the UBCâ Rapid test were dichotomized for further analysis. Statistical analysis The median cytokeratin concentrations of the UBCâ Rapid test were 8.7 µg/L in LG-BC, and 32.1 µg/L in HG-BC The mean concentrations of the UBCâ Rapid test were excluding CIS, 29.3 µg/L in patients with concomitant CIS compared using Welch’s two-sided t-test. The diagnostic and 19.4 µg/L in patients with solitary CIS, respectively. The accuracy of the UBCâ Rapid test was evaluated by receiver- mean cytokeratin concentration of the UBCâ Rapid test was operating characteristic (ROC) curve analysis. Sensitivity, significantly higher in tumour patients (56.1 µg/L) than in specificity, positive predictive value (PPV), negative predictive non-tumour patients (14.8 µg/L; P < 0.001). Furthermore, the value (NPV), and the area under the curve (AUC) were mean cytokeratin concentration was lower in LG-BC patients calculated at the optimal cut-off identified using the maximal (35.1 µg/L) than in patients with HG-BC excluding CIS Youden index. (76.2 µg/L; P < 0.001) and lower than in patients with For a subgroup of 291 patients with primary bladder cancer concomitant CIS (87.8 µg/L; P < 0.001), but not significantly and 303 healthy controls from HELIOS Hospital Bad Saarow, different from the concentration in patients with solitary CIS University Hospital Schleswig-Holstein Campus L€ ubeck and (34.7 µg/L; P = 0.972). The mean cytokeratin concentrations Lukas Hospital Neuss (all Germany), data on established risk in patients with primary BC (61.1 µg/L; P = 0.395) were not factors were available. This subgroup was divided randomly significantly different from those in patients with recurrent into a development cohort of 320 patients and a validation BC (54.8 µg/L). The median cytokeratin concentrations were cohort of 274 patients. Exclusion of patients with HG-BC at 11.1 µg/L and 18.3 µg/L, respectively. from the development and validation group of the LG model The overall sensitivity, specificity and AUC for the and vice versa resulted in group sizes of n = 244 and n = 200 UBCâ Rapid test alone were 60.4%, 75.5% and 0.68 (95% CI for the development and validation of the LG model, 0.66–0.70), respectively. Both sensitivity and AUC were respectively, and group sizes of n = 237 and n = 216 for the higher for HG-BC excluding CIS (70.5% and 0.73, 95% CI development and validation of the HG model, respectively. 0.70–0.76) than for LG-BC (46.4% and 0.61, 95% CI 0.58– Patients with CIS were included in the HG group according 0.64). Concomitant CIS was detected with a sensitivity of to the 2004 WHO classification [20]. 85.2%, with an AUC of 0.80 (95% CI 0.76–0.85), while The cohorts were compared using analysis of variance solitary CIS was detected with a sensitivity of 68.0% and an (ANOVA) and chi-squared tests. Multivariate logistic AUC of 0.72 (95% CI 0.62–0.81). The PPVs for all patients, regression models were fitted within the development cohort. and those with LG-BC, HG-BC, concomitant and solitary CIS The association of potential risk factors, including UBCâ were 64.7%, 36.1%, 46.4%, 17.2% and 6.3%, respectively. The Rapid test results, age, smoking status and haematuria, with NPVs were 71.9%, 82.5%, 89.5%, 98.8% and 99.0%, the presence of BC was investigated, and the results were respectively. The UBCâ Rapid test detected primary BC with reported with odds ratios and 95% CIs. A backward stepwise a higher sensitivity (65.3%), specificity (79.0%) and AUC selection based on the lowest Akaike information criterion (0.72; 95% CI 0.69–0.75) than it did recurrent BC, for which was used to develop the most parsimonious and accurate the sensitivity, specificity and AUC were 54.1%, 71.1% and model. Two separate models to predict primary LG-BC and 0.63 (95% CI 0.58–0.67), respectively. There was no HG-BC were developed and depicted in the nomogram significant difference in tumour grade distribution between format. Both models were externally validated in a validation the screening group, which consisted of patients with primary cohort. External validation focused on AUC, calibration BC and healthy controls, and the surveillance group, which Ó 2022 The Authors. 756 BJU International published by John Wiley & Sons Ltd on behalf of BJU International. Table 1 Overview of study approvals and patient numbers in the respective study centres. Study centre Regional ethics committee Reference number n Tumour, n (%) Control, n (%) Excluded, n (%) University Hospital Innsbruck, Local ethics committee of the AN2016-0056; 360/4.7 75 31 (41.3) 44 (58.7) 0 (0.0) Austria Medical University of Innsbruck University Hospital Tu € bingen, Local ethics committee of the 032/2013BO2 198 61 (30.8) 137 (69.2) 0 (0.0) Germany Medical University of Tu € bingen University Hospital of Umea, Regional ethical board of the 2012-478-31M/2013-364-32M 112 35 (31.2) 66 (58.9) 11 (9.8) Sweden Ume a University Danderyd Hospital, Sweden Regional ethical board of the 2012-478-31M/2013-364-32M 31 31 (100) 0 (0.0) 0 (0.0) Ume a University Enko € ping Hospital, Sweden Regional ethical board of the 2012-478-31M/2013-364-32M 13 12 (92.3) 0 (0.0) 1 (7.7) Ume a University Uppsala University Hospital, Regional ethical board of the 2012-478-31M/2013-364-32M 120 37 (30.8) 68 (56.7) 15 (12.5) Sweden Ume a University Uppsala University Hospital, Regional ethics committee of 2017/127 96 9 (9.4) 84 (87.5) 3 (3.1) Sweden Uppsala Enko € ping Hospital, Sweden Regional ethics committee of 2017/127 85 11 (12.9) 72 (84.7) 2 (2.4) Uppsala Fundacio Puigvert, Universitat Ethics committee for 30 6 (20.0) 23 (76.7) 1 (3.3) Auto  noma de Barcelona, Investigation with medicinal Spain products (CEIm) of Fundacio  Puigvert National University Health National Healthcare Group 2017/00486 112 52 (46.4) 58 (51.8) 2 (1.8) System, Singapore Domain Specific Review Board HELIOS Hospital Bad Saarow, Local ethics committee of the AS147(bB)/2013 460 242 (52.6) 197 (42.8) 21 (4.6) Germany Landesa € rztekammer Brandenburg Charite € tsmedizin  – Universita Local ethics committee of the AS147(bB)/2013 113 47 (41.6) 59 (52.2) 7 (6.2) Berlin, Germany Landesa € rztekammer Brandenburg Rheinland Klinikum Neuss, Local ethics committee of the AS147(bB)/2013 313 111 (35.5) 178 (56.9) 24 (7.7) Germany Landesa € rztekammer Brandenburg Rheinland Klinikum Neuss, Local ethics committee of the 2019071 64 36 (56.2) 26 (40.6) 2 (3.2) Germany Landesa € rztekammer Nordrhein University Hospital Du € sseldorf, Local ethics committee of the 2019024992; 6260R (MPG §23b) 25 13 (52.0) 11 (44.0) 1 (4.0) Germany medical faculty of the Heinrich Heine University Du € sseldorf University Hospital Schleswig- Local ethics committee of the 19–214 30 30 (100) 0 (0.0) 0 (0.0) Holstein, Campus Lu € beck, University Lu€ beck Germany BJU International published by John Wiley & Sons Ltd on behalf of BJU International. Ó 2022 The Authors. 757 Nomograms including the UBCâ Rapid test 1464410x, 2022, 6, Downloaded from https://bjui-journals.onlinelibrary.wiley.com/doi/10.1111/bju.15677 by Readcube (Labtiva Inc.), Wiley Online Library on [30/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 1464410x, 2022, 6, Downloaded from https://bjui-journals.onlinelibrary.wiley.com/doi/10.1111/bju.15677 by Readcube (Labtiva Inc.), Wiley Online Library on [30/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Meisl et al. Table 2 Clinical characteristics of the patient cohort and results of the UBCâ Rapid test with a cut-off of 9.88 µg/L. All patients Tumour Primary tumour NED Healthy Development Validation P* control N 1787 763 453 456 568 320 274 Age, years Mean (SD†) 69.0 (12.8) 71.8 (10.7) 71.1 (11.0) 70.6 (10.5) 64.2 (15.5) 69.5 (12.2) 69.6 (12.5) 0.95 Range 20–98 26–98 26–95 22–92 20–93 26–92 20–94 Gender, n (%) Male 1355 (76) 597 (78) 346 (76) 353 (77) 405 (71) 238 (74) 192 (70) 0.28 Female 432 (24) 166 (22) 107 (24) 103 (23) 163 (29) 82 (26) 82 (30) UBCâ Rapid Positive, n (%) 712 (40) 461 (60) 296 (65) 132 (29) 119 (21) 113 (35) 105 (38) 0.50 Negative, n (%) 1075 (60) 302 (40) 157 (35) 324 (71) 449 (79) 207 (65) 169 (62) Mean (SD), µg/L 32.4 (64.8) 56.1 (85.9) 61.1 (89.2) 15.3 (32.1) 14.4 (33.9) 36.8 (75.9) 30.7 (59.2) 0.28 Median, µg/L 7.0 15.5 18.3 5.8 5.0 6.5 5.6 Smoking status, n (%) Active 230 (13) 133 (17) 90 (20) 39 (9) 58 (10) 82 (26) 66 (24) 0.77 Former 368 (21) 208 (27) 127 (28) 61 (13) 99 (17) 116 (36) 107 (39) Never 358 (20) 133 (17) 78 (17) 77 (17) 148 (26) 122 (38) 101 (37) Missing 831 (47) 289 (38) 158 (35) 279 (61) 263 (46) 0 (0) 0 (0) Haematuria, n (%) Positive 722 (40) 447 (59) 324 (72) 91 (20) 184 (32) 170 (53) 157 (57) 0.35 Negative 665 (37) 178 (23) 96 (21) 106 (23) 381 (67) 150 (47) 117 (43) Missing 400 (22) 138 (18) 33 (7) 259 (57) 3 (1) 0 (0) 0 (0) Group, n (%) Tumour 763 (43) 453 (100) 159 (50) 132 (48) 0.78 LG 306 (17) 184 (41) 83 (26) 58 (21) 0.31 HG 308 (17) 198 (44) 62 (19) 63 (23) Concomitant CIS 61 (3) 38 (8) 14 (4) 11 (4) Solitary CIS 25 (1) 2 (0.4) 0 (0) 0 (0) Missing 63 (4) 31 (7) 0 (0) 0 (0) NED 456 (26) 0 (0) 0 (0) 0 (0) Healthy control 568 (32) 0 (0) 161 (50) 142 (52) TNM Tumour 763 (43) 453 (100) 159 (50) 132 (48) Ta 431 (24) 234 (52) 88 (28) 62 (23) 0.60 T1 158 (9) 102 (23) 31 (10) 35 (13) T2 126 (7) 96 (21) 34 (11) 30 (11) T3 17 (1) 14 (3) 6 (2) 5 (2) Solitary CIS 25 (1) 2 (0) 0 (0) 0 (0) Missing 6 (0) 5 (1) 0 (0) 0 (0) CIS, carcinoma in situ; HG, high-grade; LG, low-grade; NED, no evidence of disease. *P values regarding the comparison of the development and validation subgroup using analysis of variance (ANOVA) and chi-squared tests. †Standard deviation. consisted of patients with recurrent BC and patients with small number of patients and this was therefore excluded as a NED (P < 0.001). The PPV and NPV were 71.3% and 74.1%, predictor for the model. in the screening group, and 44.5% and 78.3% in the Based on the multivariate logistic regression models, we surveillance group, respectively. developed two nomograms to predict the risk of primary LG- To identify independent risk factors for the development of a BC and primary HG-BC (Fig. 1). The nomograms showed predictive model, univariate logistic regression analysis was robust discrimination after external validation, with an AUC performed for the predictors age, gender, UBCâ Rapid test of 0.79 (95% CI 0.72–0.87) for the LG model and 0.95 results, smoking status and haematuria in a cohort of 320 (95% CI 0.92–0.98) for the HG model. Brier scores of 0.15 patients with documented risk factors (Table 3). Gender and age for the model for LG-BC and 0.09 for the model for HG-BC showed no significant association with the presence of BC in the proved that both models were well calibrated. The calibration LG model, and gender showed no association in the HG model. plots shown in Fig. 2 indicate a constant overestimation of Backward selection showed that multivariate models that the risk for BC in both models. included age, UBCâ Rapid test results, smoking and haematuria The decision curve analysis results (Fig. 3) showed that the had the lowest Akaike information criteria (Table 3). Cytology is decision to investigate further based on the nomograms a highly specific predictor of the presence of bladder cancer [9]. provided a net benefit over a decision based solely on the However, due to the heterogeneity in study protocols across results of the UBCâ Rapid test across all low- and study centres, data on this risk factor were only available for a Ó 2022 The Authors. 758 BJU International published by John Wiley & Sons Ltd on behalf of BJU International. 1464410x, 2022, 6, Downloaded from https://bjui-journals.onlinelibrary.wiley.com/doi/10.1111/bju.15677 by Readcube (Labtiva Inc.), Wiley Online Library on [30/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Nomograms including the UBCâ Rapid test intermediate-risk thresholds, which are the most relevant to in patients without tumours (56.1 µg/L in patients with the decision on further treatment. This was true for tumours, 15.3 µg/L in patients with NED, and 14.4 µg/L in thresholds below 30% for the LG model and for risk the healthy control group). Studies analysing the UBCâ Rapid thresholds below 35% for the HG model. test reported a cut-off of 6.7–12.0 µg/L, with a sensitivity of 56%–71% and a specificity of 61%–96% [13–19]. These data The results of the cut-off analysis including sensitivity, are consistent with the sensitivity of 60.4% and the specificity specificity, PPV and NPV for various cut-off values are of 75.5% calculated in this study. The sensitivity was 70.5% presented in Table 4. The use of a threshold risk of 15% would for the detection of HG tumours, 85.2% for concomitant CIS result in no recommendation to pursue further diagnostics for and 68.0% percent for solitary CIS, whereas it was 46.4% for LG-BC and HG-BC in 51 (25.5%) and 104 (48.1%) patients in LG-BC. As a parameter of diagnostic accuracy, the AUC of the validation group, respectively, and BC would be missed in 0.68 (95% CI 0.66–0.70) determined in this study is at the five cases (2.5%) and three cases (1.4%), respectively. lower end of the range of 0.68–0.84 previously reported [13– Conversely, further diagnostics for LG-BC and HG-BC would 19]. With a cut-off value of 9.88 µg/L, the UBCâ Rapid test be recommended in 149 (74.5%) and 112 patients (51.9%), of is better suited to rule out BC than to detect it. whom 53 (26.5%) and 71 (32.9%) would be correctly identified as tumour patients, respectively. For patients with HG non- The cytokeratin concentrations determined by the UBCâ muscle-invasive BC (NMIBC) and HG muscle-invasive BC Rapid test were significantly higher in patients with (MIBC), no further diagnostics would be recommended in 102 concomitant CIS (P < 0.001) and HG-BC (P < 0.001) than in (56.4%) and 103 patients (58.2%), among which one case patients with LG-BC, which is consistent with previous (0.6%) and two cases (1.1%) would falsely be classified as reports [13,18]. Although the UBCâ Rapid test can predict negative, respectively. Further diagnostics would be advised in HG-BC and CIS reliably, similar to other POC tests for BC 79 (43.6%) and 74 patients (41.8%), of whom 38 (21.0%) and 33 such as the bladder tumour antigen (BTA) and nuclear (18.6%) would have BC, respectively. This equals a sensitivity of matrix protein 22 (NMP22) assays, it is of limited diagnostic 97.4% and 94.3%, respectively, and a specificity of 71.1%. value for LG-BC and lacks the specificity of urine cytology [11]. With an AUC of 0.72 (95% CI 0.69–0.75) for the detection of primary BC found in this study, the UBCâ Rapid Discussion test is better suited for the screening of healthy patients than Decisions regarding the best clinical management of patients for the surveillance of patients at risk of recurrence, where at risk of BC can be challenging. This analysis aimed to the AUC was 0.65 (95% CI 0.60–0.70). Due to a similar evaluate the clinical usefulness of the quantitative UBCâ tumour grade distribution (P < 0.001) between the screening Rapid test for the detection of BC. We developed two group and the surveillance group, tumour grade was excluded nomograms that describe the patient-specific risk for primary as a confounding factor for this effect. LG-BC and HG-BC. Based on these findings, we have developed what is, to our The cytokeratin concentration measured with the UBCâ knowledge, the first nomogram for assessing bladder cancer Rapid test was significantly higher in patients with BC than risk in healthy patients. In addition to the UBCâ Rapid test Table 3 Univariate und multivariate logistic regression analyses for the nomogram development subgroup analysing associations with primary low-grade bladder cancer (BC; n = 245) and primary high-grade BC (n = 236). Factor Univariate model Multivariate model OR 95% CI P OR 95% CI P LG-BC Age 1.02 0.99–1.04 0.177 1.03 1.00–1.05 0.066 Gender (female) 1.09 0.59–1.96 0.786 UBCâ Rapid 7.39 3.91–14.47 0.000 5.39 2.66–11.25 0.000 Smoking (active) 2.71 1.31–5.62 0.007 3.73 1.53–9.38 0.004 Smoking (former) 2.09 1.14–3.91 0.018 3.46 1.64–7.69 0.002 Haematuria 5.06 2.89–9.06 0.000 5.01 2.61–9.91 0.000 HG-BC Age 1.04 1.01–1.07 0.003 1.08 1.04–1.14 0.001 Gender (female) 0.82 0.42–1.53 0.537 UBCâ Rapid 20.81 10.55–43.14 0.000 12.36 5.01–33.34 0.000 Smoking (active) 6.15 3.01–13.05 0.000 12.57 3.75–48.62 0.000 Smoking (former) 2.14 1.06–4.39 0.035 5.99 1.91–21.03 0.003 Haematuria 30.07 13.12–81.97 0.000 41.31 13.56–157.78 0.000 BC, bladder cancer; HG, high-grade; LG, low-grade; OR, odds ratio. Ó 2022 The Authors. BJU International published by John Wiley & Sons Ltd on behalf of BJU International. 759 1464410x, 2022, 6, Downloaded from https://bjui-journals.onlinelibrary.wiley.com/doi/10.1111/bju.15677 by Readcube (Labtiva Inc.), Wiley Online Library on [30/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Meisl et al. Fig. 1 Nomograms predicting the probability of low-grade bladder cancer (LG-BC) (A) and high-grade bladder cancer (HG-BC) (B) using age, UBCâ Rapid test, smoking status and haematuria. To calculate the likelihood of each type of bladder cancer (BC) in a patient, the score for each risk factor is measured on the point scale axis. The total score is calculated by adding up each individual score. The probability of BC is estimated by projecting the total score onto the lower total score scale. (A) 0 10 20 30 40 50 60 70 80 90 100 Points Age (years) 25 30 35 40 45 50 55 60 65 70 75 80 85 90 pos UBC Rapid neg former Smoker never active pos Haematuria neg Total Points 0 50 100 150 200 250 300 350 400 Risk of LG-BC 0.05 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 (B) 0 10 20 30 40 50 60 70 80 90 100 Points Age (years) 30 35 40 45 50 55 60 65 70 75 80 85 90 95 pos UBC Rapid neg former Smoker never active pos Haematuria neg Total Points 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 Risk of HG-BC 0.001 0.01 0.05 0.1 0.25 0.5 0.75 0.9 0.95 0.99 results, the logistic regression models identified age, smoking nomograms. In the models developed in this study, patient status and haematuria as factors predicting the risk of age was the biggest contributor to the risk of BC. The role of primary BC. Although BC is four times more common in the factor age in risk assessment was less important in LG-BC men [1], gender did not play a crucial role in our than in HG-BC. Ó 2022 The Authors. 760 BJU International published by John Wiley & Sons Ltd on behalf of BJU International. 1464410x, 2022, 6, Downloaded from https://bjui-journals.onlinelibrary.wiley.com/doi/10.1111/bju.15677 by Readcube (Labtiva Inc.), Wiley Online Library on [30/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Nomograms including the UBCâ Rapid test Fig. 2 Calibration plot to estimate the performance of the nomogram for Fig. 3 Decision curve analysis comparing the net benefit of using the the prediction of low-grade bladder cancer (LG-BC) (A) and high-grade developed nomogram for low-grade bladder cancer (LG-BC) (A) and bladder cancer (HG-BC) (B). Ideal prediction corresponds to a slope of 1. high-grade bladder cancer (HG-BC) (B) for decision making (yellow line) The solid line represents the logistic calibration curve. The vertical lines at with using exclusively the results of UBCâ Rapid test (green line) and with the bottom indicate the risk distribution of the patients in the validation the strategy of assuming no patients have LG-BC or HG-BC (black line) or cohort. all patients have LG-BC or HG-BC (blue line), respectively. (A) (A) 1.0 Treat all 0.5 Treat none Treat based on UBC Rapid Test Treat based on nomogram 0.8 Actual Probability for LG-BC Net benefit 0.3 0.6 0.1 -0.1 0.4 0.0 0.2 0.4 0.6 0.8 1.0 Ideal Threshold probability for LG-BC (%) 0.2 Logistic calibration (B) Treat all 0.5 Treat none 0.0 Treat based on UBC Rapid Test Treat based on nomogram Net benefit 0.0 0.2 0.4 0.6 0.8 1.0 0.3 0.1 (B) 1.0 -0.1 Actual Probability for HG-BC 0.8 0.0 0.2 0.4 0.6 0.8 1.0 Threshold probability for HG-BC (%) 0.6 10, interleukin-8, vascular endothelial growth factor, 0.4 angiogenin and syndecan) from 686 patients to obtain a model for BC prediction with an AUC of 0.89. The AUC Ideal values of the demographic model and the biomarker model 0.2 Logistic calibration alone were 0.81 and 0.84, respectively. Their observations suggest that including a large variety of factors in a model 0.0 can increase the predictive power of a nomogram. The 0.0 0.2 0.4 0.6 0.8 1.0 major advantage of our nomogram, which focuses on Predicted Probability for HG-BC established and readily available risk factors, is its clinical applicability. New laboratory techniques have led to significant advances in To assess the overall probability of BC, multiple models the evaluation of urinary tumour RNA as diagnostic markers. have been developed [22–24]. A multivariate logistic This allowed for the development of a nomogram based on regression model, including age, gender, smoking status the expression of a cluster of urinary microRNA consisting of and degree of haematuria in 1182 patients showed an let-7c-5p miRNA, miR-99a-5p, and 125b-5p and the AUC of 0.74, while another nomogram including age, sex, European Organization for Research and Treatment of race, degree of haematuria and smoking status in 4178 Cancer (EORTC) score for progression to predict the patients was able to predict the presence of BC with an probability of progression-free survival. The nomogram helps AUC of 0.79 [22,23]. The present results support the to identify patients at risk of progression with an AUC of notion that asymptomatic haematuria in combination with 0.87 [25]. In addition to RNA, the analysis of epigenetic and demographic markers plays an important role in BC risk genetic alterations such as DNA methylation, histone tail assessment. modifications, and DNA mutations offers a promising new Huang et al. [24] proposed another nomogram combining approach for the diagnosis of bladder cancer. However, novel demographic factors (age, race and smoking habits) with markers, such as the established tests, are often limited by biomarker data (matrix metalloproteinase [MMP]-9, MMP- low specificity [11]. Ó 2022 The Authors. BJU International published by John Wiley & Sons Ltd on behalf of BJU International. 761 1464410x, 2022, 6, Downloaded from https://bjui-journals.onlinelibrary.wiley.com/doi/10.1111/bju.15677 by Readcube (Labtiva Inc.), Wiley Online Library on [30/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Meisl et al. Table 4 Sensitivity, specificity, and positive and negative predictive values of 0.52 and 0.62, respectively [30]. The development of for the prediction of low-grade bladder cancer (LG-BC) and high-grade bladder cancer (HG-BC) at different thresholds for the nomogram-derived models with high accuracy for the prediction of the risk of risk value for LG-BC and HG-BC. tumour recurrence was not possible based on our dataset. Threshold Sensitivity Specificity PPV NPV The main limitation of this study was the small number of (%) (%) (%) (%) (%) patients involved in the development of the nomograms, LG-BC which limited both the statistical interpretation and the 5 98.3 4.9 29.7 87.5 quality of the nomograms. This can be attributed to the 10 93.1 26.8 34.2 90.5 heterogeneity of the data from the different study centres, 15 91.4 32.4 35.6 90.2 20 86.2 49.3 41.0 89.7 which were collected retrospectively. Therefore, data on 25 77.6 63.4 46.4 87.4 significant risk factors as well as urine cytology results were 30 74.1 66.9 47.8 86.4 only available in some of the study centres. However, given 35 70.7 73.2 51.9 86.0 40 70.7 78.9 57.7 86.8 its high specificity, the inclusion of urine cytology as a risk HG-BC factor in the models might further improve their 5 98.6 62.7 57.9 98.9 performance. This was shown by Cha et al. [22] in their 10 95.9 68.3 61.2 97.0 15 95.9 71.1 63.4 97.1 above-mentioned BC risk model, where adding cytology 20 94.6 71.8 63.6 96.2 results to the model significantly increased the AUC from 25 94.6 75.4 66.7 96.4 0.74 to 0.83. In addition, the influence of haematuria could 30 94.6 78.9 70.0 96.6 35 91.9 82.4 73.1 95.1 not be assessed separately for microscopic and macroscopic 40 90.5 85.2 76.1 94.5 haematuria. BC, bladder cancer; HG, high-grade; LG, low-grade; NPV, negative Because we focused on the clinical applicability of the models predictive value; PPV, positive predictive value. and because data for all risk factors may not be available for all patients, we developed the BLUCAB Indexâ, a web-based Decision curve analysis indicated the clinical usefulness of the calculator based on our dataset. The BLUCAB Indexâ can proposed nomograms. At a risk threshold of 15%, the LG and predict the probability of primary and recurrent LG-BC, HG- HG models showed a sensitivity of 91.4% and 95.9%, and a BC and BC in general for any combination of risk factors. specificity of 32.4% and 71.1%, respectively, for the detection The BLUCAB Indexâ is available for free at www.blucab- of LG-BC and HG-BC. In addition, the HG model was highly index.net. accurate in detecting HG-NMIBC as well as HG-MIBC. Due In conclusion, the UBCâ Rapid test alone has limited clinical to the high sensitivity, at this risk threshold the nomograms utility for predicting primary or recurrent BC. Two could also be used for so-called ‘reflex testing’ to increase the nomograms were developed to estimate the risk of primary accuracy of a previous test, for example, urine cytology [26]. BC in patients who have undergone the UBCâ Rapid test, However, validation of the potential benefits of this procedure including the predictors age, smoking status and haematuria. is still required. Other single urine biomarkers such as The nomograms provide a highly accurate tool for the BTAstatâ, NMP22â BladderChekâ and ImmunoCyt achieve prediction of primary LG-BC and especially primary HG-BC, sensitivity values of 64%, 58% and 78%, respectively, and and have good discrimination and calibration. The UBCâ specificity values of 77%, 88% and 78%, respectively [27]. Rapid test in combination with these predictors might Applying a risk threshold of 15% to our models, BC was not facilitate clinical decision making for patients at risk of detected in 0.6% and 1.1% of patients in the validation group, primary BC. respectively. In comparison, patients with LG-NMIBC were classified as false-negative in 2.5% of cases. Although 75% Conflicts of Interest percent of patients present with NMIBC at time of diagnosis, Johan Styrke and Amir Sherif received grants from IDL patients who are diagnosed with MIBC have a much higher Biotech, Bromma, Sweden in 2014. Thorsten H. Ecke received mortality risk due to rapid tumour progression, which is why grants from IDL Biotech, Bromma, Sweden in 2019–2021. early diagnosis is particularly important [4]. Roland Einarsson and Per-Uno Malmstr€om are scientific The models presented here offer a good complement to the advisors to IDL Biotech. Arnulf Stenzl, Rene Ritter and J€ org EORTC risk tables and CUETO (Club Urologico Espa~ nol de Hennenlotter received grants from IDL Biotech, Bromma, Tratamiento Oncologico) scoring models, which can be used Sweden in 2013. The test systems were sponsored by Concile to assess the risk of tumour recurrence and progression of GmbH, Freiburg/Breisgau, Germany, and IDL Biotech, NMIBC [28,29]. In an external validation, the EORTC model Bromma, Sweden. showed an AUC for disease recurrence and progression of 0.60 and 0.66, respectively. The CUETO model could Acknowledgment discriminate disease recurrence and progression with an AUC Open access funding enabled and organized by ProjektDEAL. Ó 2022 The Authors. 762 BJU International published by John Wiley & Sons Ltd on behalf of BJU International. 1464410x, 2022, 6, Downloaded from https://bjui-journals.onlinelibrary.wiley.com/doi/10.1111/bju.15677 by Readcube (Labtiva Inc.), Wiley Online Library on [30/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Nomograms including the UBCâ Rapid test References 19 Ritter R, Hennenlotter J, K€ uhs U et al. Evaluation of a new quantitative point-of-care test platform for urine-based detection of bladder cancer. 1 Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A. Global Urol Oncol 2014; 32(3): 337–44 cancer statistics 2018: GLOBOCAN estimates of incidence and mortality 20 Eble JL, Sauter G, Epstein JI. Pathology and Genetics of Tumours of the worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2018; 68: 394–424 Urinary System and Male Genital Organs: WHO Classification of 2 Safiri S, Kolahi A-A, Naghavi M. Global, regional and national burden of Tumours, Vol. 7, Reprint. Lyon: World Health Organization: 2004 bladder cancer and its attributable risk factors in 204 countries and 21 Sobin LH, Wittekind C, Gospodarowicz MK eds. TNM: Classification of territories, 1990–2019: a systematic analysis for the Global Burden of Malignant Tumours, 7th edn. New York: Wiley: 2009 Disease study 2019. BMJ Glob Health 2021; 6: e004128 22 Cha EK, Tirsar LA, Schwentner C et al. Accurate risk assessment of 3 van Osch FH, Jochems SH, van Schooten FJ, Bryan RT, Zeegers MP. patients with asymptomatic hematuria for the presence of bladder bancer. Quantified relations between exposure to tobacco smoking and bladder World J Urol 2012; 30: 847–52 cancer risk: a meta-analysis of 89 observational studies. Int J Epidemiol 2016; 45: 857–70 23 Matulewicz RS, Rademaker A, Meeks JJ. A simplified nomogram to assess risk of bladder cancer in patients with a new diagnosis of 4 Ramirez D, Gupta A, Canter D et al. Microscopic haematuria at time of microscopic hematuria. Urol Oncol 2020; 38: 240–6 diagnosis is associated with lower disease stage in patients with newly diagnosed bladder cancer. BJU Int 2016; 117: 783–6 24 Huang S, Kou L, Furuya H et al. A nomogram derived by combination of demographic and biomarker data improves the noninvasive evaluation 5 Linder BJ, Bass EJ, Mostafid H, Boorjian SA. Guideline of guidelines: of patients at risk for bladder cancer. Cancer Epidemiol Biomarkers Prev asymptomatic microscopic haematuria. BJU Int 2018; 121: 176–83 2016; 25: 1361–6 6 Babjuk M, Burger M, Comperat E. EAU guidelines. In Edn. Presented at 25 Spagnuolo M, Costantini M, Ferriero M et al. Urinary expression of let- the EAU Annual Congress Milan, Chapters 5.5 and 5.8. Arnhem: EAU 7c cluster as non-invasive tool to assess the risk of disease progression in Guidelines Office, 2021: 13–5. ISBN 978-94-92671-13-4 patients with high grade non-muscle invasive bladder Cancer: a pilot 7 Garcia-Perdomo HA, Jimenez-Mejias E, Lopez-Ramos H. Efficacy of study. J Exp Clin Cancer Res 2020; 39: 68 antibiotic prophylaxis in cystoscopy to prevent urinary tract infection: a 26 Schmitz-Dr€ager BJ, Droller M, Lokeshwar VB et al. Molecular markers systematic review and meta-analysis. Int Braz J Urol 2015; 41: 412–24 for bladder cancer screening, early diagnosis, and surveillance: the WHO/ 8 Burger M, Oosterlinck W, Konety B et al. ICUD-EAU International ICUD consensus. Urol Int 2015; 94: 1–24 Consultation on Bladder Cancer 2012: Non-muscle-invasive urothelial 27 Chou R, Gore JL, Buckley D et al. Urinary biomarkers for diagnosis of carcinoma of the bladder. Eur Urol 2013; 63: 36–44 bladder cancer: a systematic review and meta-analysis. Ann Intern Med 9 Lotan Y, Roehrborn CG. Sensitivity and specificity of commonly available 2015; 163: 922–31 bladder tumor markers versus cytology: results of a comprehensive 28 Fernandez-Gomez J, Madero R, Solsona E et al. Predicting nonmuscle literature review and meta-analyses. Urology 2003; 61: 109–18 invasive bladder cancer recurrence and progression in patients treated 10 Taylor J, Becher E, Steinberg GD. Update on the guideline of guidelines: with bacillus Calmette-Guerin: the CUETO scoring model. J Urol 2009; non-muscle-invasive bladder cancer. BJU Int 2020; 125: 197–205 182: 2195–203 11 Ng K, Stenzl A, Sharma A, Vasdev N. Urinary biomarkers in bladder 29 Sylvester RJ, van der Meijden AP, Oosterlinck W et al. Predicting cancer: A review of the current landscape and future directions. Urol recurrence and progression in individual patients with stage Ta T1 Oncol 2021; 39: 41–51 bladder cancer using EORTC risk tables: a combined analysis of 2596 12 Southgate J, Harnden P, Trejdosiewicz LK. Cytokeratin expression patients from seven EORTC trials. Eur Urol 2006; 49: 466–5 patterns in normal and malignant urothelium: a review of the biological 30 Xylinas E, Kent M, Kluth L et al. Accuracy of the EORTC risk tables and and diagnostic implications. Histol Histopathol 1999; 14: 657–64 of the CUETO scoring model to predict outcomes in non-muscle-invasive 13 Ecke TH, Weiß S, Stephan C et al. UBCâ Rapid Test for detection of urothelial carcinoma of the bladder. Br J Cancer 2013; 109: 1460–6 carcinoma in situ for bladder cancer. Tumour Biol 2017; 39: 1010428317701624 14 Pichler R, Tulchiner G, Fritz J, Schaefer G, Horninger W, Heidegger I. Correspondence: Christina J. Meisl, Department of Urology, Urinary UBC rapid and NMP22 test for bladder cancer surveillance in Charite - Universit€atsmedizin Berlin, Chariteplatz 1, 10117 comparison to urinary cytology: results from a prospective single-center Berlin, Germany. study. Int J Med Sci 2017; 14: 811–9 15 Gleichenhagen J, Arndt C, Casjens S et al. Evaluation of a new survivin e-mail: christina.meisl@charite.de ELISA and UBCâ rapid for the detection of bladder cancer in urine. Int J Abbreviations: AUC, area under the curve; BC, bladder Mol Sci 2018; 19: 226 16 Lu P, Cui J, Chen K et al. Diagnostic accuracy of the UBCâ Rapid Test cancer; CIS, carcinoma in situ; CUETO, Club Urologico for bladder cancer: A meta-analysis. Oncol Lett 2018; 16: 3770–8 Espa~nol de Tratamiento Oncologico; EORTC, European 17 Styrke J, Henriksson H, Ljungberg B et al. Evaluation of the diagnostic Organization for Research and Treatment of Cancer; HG, accuracy of UBCâ Rapid in bladder cancer: a Swedish multicentre study. high-grade; LG, low-grade; MIBC, muscle-invasive bladder Scand J Urol 2017; 51: 293–300 cancer; MMP, matrix metalloproteinase; NED, no evidence of 18 Ecke TH, Weiß S, Stephan C et al. UBCâRapid test-A urinary point-of- disease; NMIBC, non-muscle-invasive bladder cancer; NPV, care (POC) assay for diagnosis of bladder cancer with a focus on non- muscle invasive high-grade tumors: results of a multicenter-study. Int J negative predictive value; POC, point-of-care; PPV, positive Mol Sci 2018; 19: 3841 predictive value. Ó 2022 The Authors. BJU International published by John Wiley & Sons Ltd on behalf of BJU International. 763