Are False-Positive Urine Markers for the Detection of Bladder Carcinoma Really Wrong or DoThey PredictTumor Recurrence?

Transcription

1 European Urology European Urology 43 (2003) 146±151 Are False-Positive Urine Markers for the Detection of Bladder Carcinoma Really Wrong or DoThey PredictTumor Recurrence? Martin G. Friedrich *, Angelika Hellstern, Marieta I. Toma, Peter Hammerer, Hartwig Huland University Hospital Eppendorf, Department of Urology, University of Hamburg, Martinistraûe 52, D Hamburg, Germany Accepted 13 November 2002 Abstract Introduction and Objectives: A problem in the interpretation of noninvasive urine tests for detection of bladder carcinoma is the nding of false-positive results. Several authors have described that patients with false-positive results are at high risk for tumor recurrence or progression. Only few data are available for comparing the clinical course of patients with false-positive test results and patients with true-negative results. We studied whether patients with false-positive results of various urine test had a higher recurrence rate than patients with true-negative results. Methods: Urine samples from 61 patients without evidence of active bladder carcinoma were included. Of the 61 patients, 51 had a history of bladder cancer, and 10 underwent transurethral resection for suspect of bladder carcinoma but had negative pathologic ndings. Immunocytology (Lewis X and 486p3/12) was performed on bladder washings, and BTAstat and NMP22 were performed on urine samples. Results: During the follow-up period, 22 patients had one or more false-positive BTAstat test results, 25 patients had one or more false-positive NMP22 tests, 42 patients had at least one false-positive Lewis X test, and 11 patients had one or more false-positive 486p3/12 test. During a follow-up period of 3±39 months (median, 17.6 months) four patients expected a tumor recurrence. Among patients with false-positive urine test results 2 of 22 (9.1%, BTAstat), 2 of 25 (8%, NMP22), 4 of 42 (9.5%, Lewis X), and 3 of 11 (27.2%, 486p3/12) suffered from tumor recurrence. In contrast, among patients with true-negative test results 2 of 39 (5.2%, BTAstat), 2 of 36 (5.6%, NMP22), 0 of 18 (0%, Lewis X), 1 of 50 (2.0%, 486p3/12) had a tumor recurrence. Conclusions: Patients with a false-positive urine test result do not generally have a greater risk of tumor recurrence or progression than patients with a true-negative result. In our series, only patients with false-positive 486p3/12 test result had a higher recurrence rate. Our ndings do not justify a more aggressive adjuvant treatment or surveillance for patients with false-positive urine tests. # 2002 Elsevier Science B.V. All rights reserved. Keywords: Urine markers; False-positive results; Prognosis 1. Introduction and objectives Cystoscopy remains the gold standard in diagnosis and surveillance of bladder cancer, however, cystoscopy * Corresponding author. Present address: Norris Cancer Comprehensive Center, University of Southern California, 1441 Eastlake Avenue, NOR 7336, Los Angeles, CA , USA. Tel ; Fax: address: friedrich@uke.uni-hamburg.de (M.G. Friedrich). is an invasive procedure. Urinary cytology has been described as a sensitive tool for detection of grade III tumors and for carcinoma in situ but offers only a poor sensitivity for low-grade tumors [1]. During the last few years, several new markers have been introduced and evaluated for noninvasive detection of bladder cancer. Among them, the BTAstat test and the NMP22 test have gained widest acceptance, and they are approved by the US Food and Drug Administration [2±8]. Both tests are /02/$ ± see front matter # 2002 Elsevier Science B.V. All rights reserved. doi: /s (02)

2 M.G. Friedrich et al. / European Urology 43 (2003) 146± used to detect a tumor-speci c protein in urine. The BTAstat test is an immunoassay that uses two monoclonal antibodies that recognize two epitopes on human complement factor H-related protein. The NMP22 test is an immunoassay that detects a nuclear mitotic apparatus protein of 240 kda that is released by tumor cells at detectable concentrations [9]. Another approach for noninvasive detection of bladder cancer is an improvement in urinary cytology that uses immunoreaction with tumor-associated antigens. For these purpose several monoclonal antibodies have been described, such as 486p3/12, LAD, M344, BG7 and P-12 [10±12]. These tests detect one, or in the case of the ImmunoCyt test, a combination of several tumor associated antigens. Other new tests for the detection of bladder cancer include the detection of tumor-associated proteins such as nuclear matrix proteins (BLCA 4) [13], clotting factor-related proteins (Accu Dx-test) [14], cytokeratins in urine, genetic alterations, or hyaluronic acid and hyaluronidase (HA-Haase) [9]. Noninvasive urine tests show encouraging results, but the interpretation of false-positive test results poses a problem in their clinical use. Several authors suggest that false-positive results can be due to suspended tumor cells in the urine associated with a tumor that is not detectable macroscopically. Nevertheless, few data are available for comparing the clinical course of patients with a false-positive test with the course of patients with a true-negative test. 2. Materials and methods 2.1. Patients We obtained urine samples and bladder washing specimens from 61 patients without active bladder cancer. Among them, 78 samples were collected from 51 patients with a history of super cial bladder (pt a, n ˆ 34; pt 1, n ˆ 17; grade I, n ˆ 11; grade II, n ˆ 34; grade III, n ˆ 6) cancer. In addition, urine and bladder washing specimens were collected from 10 patients who were scheduled for cystoscopy with bladder biopsies or transurethral resection because of symptoms including hematuria, or voiding irritation suggesting bladder cancer. Forty- ve of the 51 patients with a history of super cial bladder cancer were pretreated with intravesical chemo (Mitomycin C) or immunotherapy (BCG). All specimens were collected before any manipulation. To exclude interference with in ammation or hematuria, none of the follow-up urine samples was collected earlier than 3 months after TUR-B or 1 month after intravesical instillation. Urinary tract infection was excluded by urine dipstix investigation, or in cases with suspect of urinary tract infection (leucocyturia, hematuria, proteinuria) with urine culture. Biopsies or transurethral resection were performed to determine if cystoscopy identi ed a bladder tumor or suspicious areas such as mucosal erythema or ulceration. In all patients cystoscopy or cystoscopy and bladder biopsies revealed no evidence of malignancy. Clinical follow-up of the patients with a history of bladder carcinoma was performed by of ce urologists and included cystoscopy and urine analysis every 3 months, ultrasonography of the urinary tract every 3 months, and intravenous renography once a year to exclude tumor recurrence in the upper urinary tract. Urine samples were split for separate evaluation with the NMP22 and BTAstat test. The BTAstat test was carried out immediately while the sample for the NMP22 test was buffered in the provided buffer solution and stored at 20 8C. Bladder washings were stored at 4 8C, and Cytospin preparations were performed within 24 hours BTAstat test We used the Bard TM BTAstat test. The test was performed according to the manufacturer's instructions. We placed ve drops of urine in a well of a small disposable kit. The test was considered positive if any band appeared in the specimen window. In a control window, a band had to appear as a positive control NMP22 test The NMP22 test was distributed by Matritech TM. The test was performed according to the manufacturer's instructions. Urine samples were given on an antibody-coated enzyme linked immunosorbent assay (ELISA) plate. After washing of the plate, the bound nuclear matrix proteins reacted with a digoxigenin-labeled antibody. After a second washing, the digoxigenin-labeled antibody reacted with an antidigoxigenin antibody coupled to horseradish peroxidase; the complex was then visualized with o- phenyldiamine. The antigen concentration in the urine was proportional to the intensity of the color development. The concentration of NMP22 in each urine specimen was calculated from a standard curve. A NMP concentration of 10 U/ml was used as a threshold for a positive result Immunocytology for the Lewis X antigen and for 486p3/12 Immunocytology was performed with monoclonal BG7 IgM antibody (murine/mouse) for detection of the Lewis X antigen. Immunocytology against 486p3/12 was carried out with an IgM antibody directed against a membrane-bound glycoprotein. We used the technique as described earlier [10]. The two investigations were carried out with the same protocol. Lyophylized cytocentrifuged preparations were xed in acetone for 10 minutes and then reacted with 70 ml of diluted antibody (1:12 in phosphate buffered saline: bovine serum albumin) for 60 minutes at room temperature. Controls for nonspeci c binding were performed with normal mouse serum (70 ml, 60 minutes). Afterward, the specimens were washed twice with Tris buffer (ph, 8.7), and then incubated with alkaline phosphatase conjugated goat anti-mouse IgM (70 ml, 60 minutes) as a second antibody. After two washings with the Tris buffer, endogenous peroxidase was blocked with levamisol. The alkaline phosphatase reaction was visualized with naphtol AS biphosphonate. The cell preparations were counterstained with hemalaun and mounted. Tumor cell preparations that stained positive for 486p3/12 or BG7 were used as positive controls. In addition, leukocyte staining for 486p3/12 served as internal positive control for 486p3/12. Evaluation of the specimens was performed by counting 100 cells. Specimens with fewer than 50 cells were omitted from evaluation. The cutoff levels were taken from the literature as described by Pode et al. for the Lewis X test [12] and as described earlier by our group for the 486p3/12 [10]. Slides showing 5% or more positive cells for Lewis X, or 30% or more positive cells for 486p3/12 were considered positive.

3 148 M.G. Friedrich et al. / European Urology 43 (2003) 146± Statistical analysis The Fisher's exact test was used to investigate statistical differences between two groups. The p < 0:05 were considered to re ect signi cant differences between two groups. 3. Results 3.1. Clinical follow-up Median follow-up was 17 months (3±39 months). Among the patients with a history of bladder cancer, four of 51 (7.8%) patients had histologically proven bladder cancer recurrence, and none experienced tumor progression to a higher tumor stage. Among the other 10 patients, no evidence of tumor disease was observed False-positive results of the single urine tests BTAstat test The BTAstat test was performed on 81 samples. It produced 25 false-positive results (30.9%). Of the 61 patients, 22 (36.1%) had at least one false-positive BTAstat result NMP22 test The NMP22 test was performed on 81 samples. It produced 30 false-positive results (37%). Of the 61 patients, 25 (41%) had at least one false-positive NMP22 result Immunocytology with mab 486p3/12 Immunocytology with mab 486p3/12 was performed on 81 samples. It produced 13 false-positive results (16.0%). Of the 61 patients, 11 (18%) had at least one false-positive 486p3/12 result. Immunocytology with mab 486p3/12 showed signi cantly fewer false-positive results than the other three tests ( p < 0:05) Immunocytology with the mab BG7 against the Lewis X antigen Immunocytology with the mab BG7 was performed on 81 samples. It produced 56 false-positive results (69.1%). Of the 61 patients, 42 (69.0%) had at least one false-positive result. Immunocytology with mab BG7 showed signi cantly more false-positive results than the other three tests (p < 0:05) Combination of false positive urine tests Of the 22 patients that a false-positive BTAstat test in the follow-up, 2 patients had a more than one falsepositive BTAstat test. Of the 25 patients with at least one false-positive NMP22 test, 4 patients had a falsepositive test result for more than one time in the followup, 12/43 patients had more than one false-positive Table 1 Frequency of more than one false-positive test in the follow-up n False-positive tests BTAstat NMP p3/ Lewis X Lewis X test during the follow-up, and two patients had more than one false-positive 486p3/12 tests in the follow-up (Table 1). Among the four tests 33 of 81 (40.7%) samples (40.7%) yielded a false-positive result in only one test at the same investigation. Thirty- ve samples (43.2%) yielded a false-positive test result in more than one test. In detail, 27/81 samples (33.3%) showed false-positive results in two different tests, 7/81 samples (8.6%) in three different tests, one sample (1.2%) revealed a false-positive result in all four tests Clinical relevance of false-positive results Among the 51 patients with a history of bladder carcinoma, 4 patients (7.8%) experienced tumor recurrence after a median follow-up of 17 months (3±39 months). Among the patients with a false-positive test 2 of 22 (9.1%) (BTAstat test), 2 of 25 (8%) (NMP22 test), 4 of 42 (9.5%) patients (Lewis X test), and 3 of 11 (27.3%) patients (486p3/12) experienced tumor recurrence. Among them, none expected tumor progression. In contrast among the patients with true-negative results 2 of 39 (5.1%) (BTAstat test), 2 of 36 (5.6%) (NMP22 test), 0 of 18 (0%) (Lewis X test), and 1 of 50 (2.0%) (486p3/12) experienced tumor recurrence (Table 2). Only for the 486p3/12 test did patients with false-positive results experience signi cantly more frequent tumor recurrences ( p < 0:05) than patients with true-negative results (Table 2). There was no relationship between the frequency of false-positive tests during the follow-up or between Table 2 Frequency of tumor recurrences in patients with false negative test results and with patients with true-negative results False-positive True-negative p n Recurrence n (%) n Recurrence n (%) BTAstat 22 2 (9.1) 39 2 (5.1) n.s. NMP (8.0) 36 2 (5.6) n.s. Lewis X 42 4 (9.5) 18 0 (0) n.s. 486p3/ (27.3) 50 1 (2) <0.05

4 M.G. Friedrich et al. / European Urology 43 (2003) 146± a combination of more than one false-positive test at one investigation and the occurrence of a tumor recurrence. 4. Discussion There is a clear need for noninvasive procedures for detection of urothelial carcinoma of the bladder, whether at initial diagnosis or during follow-up. In the last few years, several new tests have been developed for noninvasive diagnosis. Among them, the NMP22 and the BTAstat tests were shown to offer superior sensitivity to urinary cytology [3±8]. The BTAstat test offers a sensitivity of 57%±83%, the speci city is described to be 46%±73%. For the NMP22 test the sensitivity has been described between 47% and 100%, and the speci city between 60% and 70% [9]. In contrast to these tests, that detect a tumorspeci c protein in urine, the Lewis X test and the 486p3/12 are based on a modi cation of the urine cytology with immunostaining against tumor-associated antigens. The Lewis X test has been described to offer a sensitivity of 79.8%±95.1% and a speci city of 86.4% [12]. The sensitivity and speci city for the 486p3/12 was described to be 77.8%±91.8%, and 66.6%, respectively [10,15]. However, the clinical use of urine tests is limited by their failure to detect small papillary and low grade lesions [16]. Another problem in the interpretation of noninvasive test results is the occurrence of false-positive results. There is always the fear that these results are positive because of suspended cells from an overlooked or macroscopically undetectable tumor. Recent publications reported a frequency of false-positive urine test results between 16.7% and 30.5% [17±20]. Themain reason for false-positive urine markers has been claimed to be caused by in ammation, previous intravesical instillation with chemo- or immunotherapy, benign prostate hyperplasia (BPH), presence or history of a foreign body, or stone disease [20,21]. Despite those nonmalignant reasons for false-positive results, several authors have stated that false-positive results can be due to tumors than cannot be visualized cystoscopically, including carcinoma in situ, or to tumors of the upper urinary tract. Therefore, it has been stated that patients with a false-positive urine test results are at a high risk for tumor recurrence or even progression [8]. This was rst described by Sarosdy et al., who found a tumor recurrence rate of about 30% in patients with false-positive BTA test results [22]. Also Pode et al. discussed that a false-positive BTAstat test result predicts tumor recurrence [4]. More recently, Lokeshwar et al. noted that almost half of the patients with false-positive urine test result (HA/ Haase-test) suffered tumor recurrence [23]. Using uorescence in situ hybridization (FISH) with probes for chromosomes 9 and 17, Ishiwata et al. described that 7 of 13 patients with positive FISH results after transurethral resection had tumor recurrences [24]. In contrast to these authors who looked only at patients with false-positive tests, we compared the recurrence rate of the latter with that of patients with true-negative tests. We found no statistically signi cant difference in frequency of tumor recurrences between the falsepositive group and the true-negative group with respect to the BTAstat test, the NMP22 test or the immunocytology with mab BG7. Only patients with false-positive immunocytology against 486p3/12 had signi cantly more frequent tumor recurrence than patients with a true-negative result. This may indicate that patients with a false-positive 486p3/12 are at an increased risk for tumor recurrence, otherwise the small number of patients with a false-positive 486p3/12 is not large enough to draw clinical conclusions out of these ndings. Our data show that the prognostic value of falsepositive urine test results must be discussed carefully. It has been shown that the results of most urine tests can be in uenced by urinary tract infections, BPH, previous intravesical instillation therapy, or stone disease. In addition, the sensitivity of urine tests is clearly correlated with tumor size [16]. It is therefore dif cult to understand why patients with minimal tumor load (macroscopically invisible tumors) would have socalled ``true±false-positive'' urine test results. It seems that only tests with high speci city may detect suspended tumor cells and thus may predict tumor recurrence. 5. Conclusions Our data show that patients with false-positive BTAstat, NMP22, or Lewis X test do not have a higher risk for tumor recurrence than patients with true-negative tests. In our series, only patients with false-positive 486p3/12 results had signi cantly more recurrences than patients with true-negative results. Considering the small number of patients with a false-positive 486p3/12, we conclude that a false-positive urine test in patients with a history of super cial bladder carcinoma does not justify more aggressive adjuvant therapy or surveillance.

5 150 M.G. Friedrich et al. / European Urology 43 (2003) 146±151 References [1] Badalament RA, Fair WR, Whitmore WF, Melamed MR. The relative value of cytometry and cytology in the management of bladder cancer. Semin Urol 1988;6:22±30. [2] Leyh H, Marberger M, Conort P, Sternberg C, Pansadoro V, Pagano F, et al. Comparison of the BTAstat test with voided urine and bladder wash cytology in diagnosis and monitoring of bladder cancer. Eur Urol 1999;35:52±6. [3] Leyh H, Mazeman E. Bard BTA test compared with voided urine cytology in the diagnosis of recurrent bladder cancer. Eur Urol 1997; 32:425±8. [4] Pode D, Shapiro A, Wald M, Nativ O, Laufer M, Kaver I. Noninvasive detection of bladder cancer with the BTAstat test. J Urol 1999;161:443±6. [5] Schamhart DH, de Reijke TM, van der Poel HG, Witjes JA, de Boer E, Kurth K, Schalken JA. The Bard BTA test: Its mode of action, sensitivity and speci city, compared to cytology of voided urine, in the diagnosis of super cial bladder cancer. Eur Urol 1998;34:99±106. [6] Ellis WJ, Blumenstein BA, Ishak LM, En eld DL. Clinical evaluation if the BTA TRAK assay and comparison to voided urine cytology and the Bard BTA test in patients with recurrent bladder tumors. The Multi Center Study Group. Urology 1997;50:882±7. [7] Chautard D, Daver A, Bocquillon V, Verielle V, Colls P, Bertrand G, Soret JY. Comparison of the BTA TRAK test with voided urine cytology in the diagnosis and follow up of bladder tumors. Eur Urol 2000;38:686±90. [8] Serretta V, Pomara G, Rizzo I, Esposito E. Urinary BTA-stat, BTA Trak and NMP22 in surveillance after TUR of recurrent super cial urothelial carcinoma of the bladder. Eur Urol 2000;38:419±25. [9] Lokeshwar VB, Soloway MS. Current bladder tumor tests: Does their projected utility ful l clinical necessity? J Urol 2001;165:1067±77. [10] Klaen R, Huland E, Baisch H, Huland H. Sensitivity of urinary quantitative immunocytology with monoclonal antibody 486p3/12 in 241 unselected patients with bladder carcinoma. J Urol 1991;145: 495±7. [11] Mian C, Pycha A, Wiener H, Haitel A, Lodde M, Marberger M. Immunocyt: A new tool for detection of urothelial carcinoma of the urinary tract. J Urol 1999;161:1486. [12] Pode D, Golijanin D, Sherman Y, Lebensart P, Shapiro A. Immunostaining of Lewis X in cells from voided urine. J Urol 1998;159:389±92. [13] Konety BR, Nguyen TS, Breenes G, Sholder A, Lewis N, Bastacky S, et al. Clinical usefulness of the novel marker BLCA-4 for the detection of bladder cancer. J Urol 2000;164:634±9. [14] Topsakal M, Karadeniz T, Anac M, DoÈnmezer S, Besisik A. Assessment of brin± brinogen degradation products (Accu Dx) test in bladder cancer patients. Eur Urol A 2000;37:386. [15] Sautner D, Saeger W. Urine examination for tumor cells using quantitative immunochemistry. Antibodies to a CEA-like antigen. Fortschr Med 1995;113:207±9. [16] Boman H, Hedelin H, HolmaÈng S. Four bladder tumor markers have a disappointing sensitivity for small size and low grade recurrence. J Urol 2002;167:80±3. [17] Raitanen M-P, Martilla T, Nurmi M, Ala-Opas M, Nieminen P, Aine R, Teuvo LJ. Human complement factor H related protein test for monitoring bladder cancer. J Urol 2001;165:374±7. [18] GutieÂrrez Banos JL, Henar Rebollo Rodrigo M, Antolin Juarez FM, Martin Garcia B. Usefulness of the BTAstat test for the diagnosis of bladder cancer. Urology 2001;57:685±9. [19] Giannopoulos A, Manousakas T, Gounari A, Constantinides C, Choremi-Papadopoulou H, Dimopoulos C. Comparative evaluation of the diagnostic performance of the BTAstat test. J Urol 2001;166: 470±5. [20] Raitanen M-P, Kaasinen E, Lukkarinen O, Kauppinen R, Viitanen J, Liukkonen T. Tammela: Analysis of false-positive BTAstat test results in patients followed up for bladder cancer. Urology 2001;57: 680±4. [21] Ponsky LE, Shashikala S, Pandrangi L, Kedia S, Nelson D, Agarwal A, Zippe CD. Screening and monitoring for bladder cancer: Re ning the use of NMP22n. J Urol 2001;166:75±8. [22] Sarosdy MF, devere White RW, Soloway MS, Shein eld J, Hudson MA, Schellhammer PF, et al. Results of a multicenter trial using the BTA test to monitor for and diagnose recurrent bladder cancer. J Urol 1995;154:379±83. [23] Lokeshwar VB, Hautmann S, Schroeder GL, Posey T, Soloway MS. HA-Hase test: Non-invasive accurate urine test for monitoring bladder cancer recurrence. Eur Urol A 2000;37:367. [24] Ishiwata S, Takahashi S, Homma Y, Tanaka Y, Kameyama S, Hosaka Y, et al. Noninvasive detection and prediction of bladder cancer by urorescence in situ hybridisation analysis of exfoliated cells in voided urine. Urology 2001;57:811±5. Editorial Comment P.-U. MalmstroÈm, Rudbeck Laboratory, Uppsala University, Uppsala, Sweden Urinary bladder carcinoma is described as the most prevalent form of cancer. This is mainly due to two factors typical of super cial disease. The recommended frequent follow-up and the duration of this because of low risk of mortality due to the disease. Approximately one-fourth of those investigated will have a recurrence [1] and these are logically of smaller size than primary tumors [2]. Earlier cytology together with cystoscopy were the mainstay for diagnosis. Recently, many new urinary markers have been proposed as adjunct to cytology or even as replacement. The later instigated by reports of very low sensitivity for routine cytology assessments. However none of the new tests have been proven as replacement of cystocopy. Efforts to improve cytology includes quantitative immunocytochemistry, DNA-image-cytometry and network-based digitized cell image methods. One example is ImmunoCyt (ucyt) that have been extensively tested with generally higher sensitivity than cytology in grades 1 and 2 tumors. The authors in this article analyze an important problem with these new tests, i.e. false-positivity. This when the diagnosis is evaluated at the same time as the test. A false-positive test could also be a signal of microscopic disease not apparent to the naked eye, that later evolves into a visible recurrence. The

6 M.G. Friedrich et al. / European Urology 43 (2003) 146± conclusion of their analysis was that generally these test are true±false-positive and not associated with an increased risk of tumor recurrence and progression. This in contrast to a recent report on screening for the disease in a cohort of Chinese workers exposed to benzidine [3]. They were prospectively analyzed with biomarkers (DNA 5CER, G-actin, p300) shed into urine, to assess the possibilities to stratify individuals for their risk of bladder cancer. Conventional screening with cytology and hematuria testing was performed as a control. Those in the high-risk group were also monitored by biannual cystoscopy. Twenty-eight bladder cancers were diagnosed in exposed workers and two in non-exposed workers. Two of the biomarkers (DNA 5CERand p300) were good markers of individual risk and predicted disease 15±33 months in advance while the conventional tests and one biomarker performed poorly. Obviously further insight into these issues are warranted. One other important question is how early recurring disease have to be diagnosed. The well-known biological heterogeneity of bladder carcionama is crucial in this decision. In the new WHO grading system from 1999 categories like papillary neoplasms of low malignant potential are de ned [4]. These represent one end of the biological spectrum of tumors that obviously does not need early diagnosis. References [1] Raitanen MP, Aine RA, Kaasinen ES, Liukkonen TJ, Kylmala TM, Huhtala H, Tammela TL, Finnbladder Group. Suspicious urine cytology (class III) in patients with bladder cancer: Should it be considered as negative or positive? Scand J Urol Nephrol 2002; 36(3):213±7. [2] Boman H, Hedelin H, Holmang S. Four bladder tumor markers have a disappointingly low sensitivity for small size and low grade recurrence. J Urol 2002;167(1):80±3. [3] Hemstreet 3rd GP, Yin S, Ma Z, et al. Biomarker risk assessment and bladder cancer detection in a cohort exposed to benzidine. J Natl Cancer Institute 2001;93(6):427±36 [see comments]. [4] Mosto FK, Davis Jr, CJ, Sesterhenn IA. Histological typing of urinary bladder tumours. World Healh Organization: International histological classi cation of tumours. Berlin: Springer; 1999.