Int Arch Occup Environ Health DOI 10.1007/s00420-013-0916-3
ORIGINAL ARTICLE
Screening for bladder cancer with urinary tumor markers in chemical workers with exposure to aromatic amines Beate Pesch • Dirk Taeger • Georg Johnen • Katarzyna Gawrych • Nadine Bonberg • Christian Schwentner • Harald Wellha¨ußer • Matthias Kluckert • Gabriele Leng • Michael Nasterlack • Yair Lotan • Arnulf Stenzl • Thomas Bru¨ning • the UroScreen Study Group
Received: 31 May 2013 / Accepted: 2 October 2013 Ó Springer-Verlag Berlin Heidelberg 2013
Abstract Purpose To validate urinary markers for the early detection of bladder cancer (BC) in chemical workers. Methods UroScreen was conducted as a validation study for tumor markers within the frame of a health surveillance program of the German Social Accident Insurance for active or retired workers with former exposure to aromatic amines. From 2003 to 2010, 1,609 men took part in voluntary annual screens. Cytology, the quantitative NMP22Ò assay, and UroVysionTM were applied to 7,091 urine samples. Results Fifteen out of 21 tumors were detected following test positivity. The UroVysion/NMP22 panel detected 14 out of 21 tumors versus 8 tumors with cytology alone Beate Pesch and Dirk Taeger have equally contributed to this article. Please refer the ‘‘Appendix’’ section for UroScreen Study Group.
(sensitivity 66.7 vs. 44.4 %, specificity 94.5 vs. 98.5 %). The sensitivity of the panel increased to 85.7 % in samples collected B12 months before diagnosis and when papillomas were excluded, compared to 58.3 % with cytology. About 3 % of NMP22 tests were false-positive. UroVysion results overlapped with cytology due to the preselection of atypical cells. NMP22 was less and UroVysion more frequently positive in diluted urine samples. Leukocytes confounded NMP22 but not UroVysion. The low incidence of BC in this study population yielded low positive predictive values of the markers and high costs per tumor detected with screening. Conclusions UroVysion in combination with NMP22 detected more cases than cytology alone, at the expense of a lower specificity. High costs per detected case resulted from a lower BC incidence than in the past when levels of occupational exposure to aromatic amines were higher. Currently, it cannot be recommended to apply these
B. Pesch (&) D. Taeger G. Johnen K. Gawrych N. Bonberg M. Kluckert T. Bru¨ning Institute for Prevention and Occupational Medicine of the German Social Accident Insurance, Institute of the RuhrUniversita¨t Bochum (IPA), Buerkle-de-la-Camp-Platz 1, 44789 Bochum, Germany e-mail: [email protected]
G. Leng Department of Health Protection and Occupational Safety, Currenta GmbH & Co. OHG, Leverkusen, Germany
N. Bonberg Protein Research Unit Ruhr within Europe (PURE), Ruhr University of Bochum, Bochum, Germany
Y. Lotan Department of Urology, University of Texas Southwestern Medical Center, Dallas, TX, USA
M. Nasterlack Occupational Medicine and Health Protection, BASF SE, Ludwigshafen, Germany
C. Schwentner A. Stenzl Institute of Urology, Eberhard Karls University, Tu¨bingen, Germany H. Wellha¨ußer M. Kluckert German Social Accident Insurance’s Institution for the Raw Materials and Chemical Industry (BG RCI), Heidelberg, Germany
123
Int Arch Occup Environ Health
markers for screening in asymptomatic workers. The increase in sensitivity is not balanced by the high costs of UroVysion and the false-positive tests of NMP22. Keywords Aromatic amines Bladder cancer Chemical workers Cytology NMP22 Screening UroVysion Validation study
Introduction Every year, bladder cancer (BC) accounts for approximately 400,000 new cases and 150,000 deaths worldwide (Jemal et al. 2011). Occupational exposure to aromatic amines has been linked to bladder cancer (Vineis and Pirastu 1997). The recognition of occupational risks dates back to the observation of bladder cancer among German aniline workers (Rehn 1895) and was attributed to exposure to 4-aminobiphenyl and 2-naphthylamine (Freudenthal et al. 1999). Both arylamines, as well as benzidine and ortho-toluidine, have been classified as human carcinogens (http://monographs.iarc.fr/). Aromatic amines are also constituents of tobacco smoke and emissions from hot tar applications or coke production, which are known to entail an increased bladder cancer risk. The International Labour Organization (ILO) has recommended health surveillance for workers with exposure to occupational carcinogens. The German Social Accident Insurance (DGUV) has established a surveillance program for workers with former exposure to carcinogenic aromatic amines aimed at the early detection of BC (Nasterlack et al. 2001). The early detection of BC with non-invasive tumor markers holds, in principle, great promise (Larre et al. 2013). Most patients with BC are diagnosed after blood is detected in the urine. Urethrocystoscopy, aided by voided urine cytology, is the gold standard for BC diagnosis. However, this invasive method is not applicable to screen asymptomatic subjects, and cytology alone suffers from a limited sensitivity (Shariat et al. 2008). Urinary markers may support the non-invasive detection of BC at early stages (Lotan et al. 2006; van Rhijn et al. 2005). Assays for the determination of nuclear matrix protein 22 (NMP22) and chromosomal instability with fluorescence in situ hybridization (FISH) have been approved for the early detection of BC (FDA 2000, 2005). BladderChek as a point-of-care test of elevated NMP22 levels has been already introduced into clinical practice. A current review revealed that no study assessed the diagnostic performance of urinary markers in the context of screening (Larre et al. 2013). The lack of a randomized controlled trial (RCT) and the shortage of prospective validation studies render performance measures for already approved markers like NMP22 inconclusive (Lokeshwar
123
et al. 2005). Factors influencing the performance of tests have not yet been sufficiently assessed. Several screening studies have been performed to detect recurrent BC in patients; but few asymptomatic at-risk cohorts have been subjected to screening, like workers exposed to aromatic amines (Hemstreet et al. 2001; Lokeshwar et al. 2002). Other studies conducted with at-risk populations were performed in clinical settings but without repeated screens (Grossman et al. 2005; Lotan et al. 2009b). UroScreen took advantage of the German surveillance program for chemical workers and offered tumor tests with urinary markers in addition to cytological examinations and to the detection of hematuria in annual screens. This prospective validation study aimed to evaluate whether a panel of tumor markers detects more primary BC cases than cytology and which factors influence the performance measures. Here, we report on the performance of NMP22 in combination with UroVysion. Detailed reports have been published elsewhere (Banek et al. 2012; Bonberg et al. 2013; Huber et al. 2012; Johnen et al. 2012; Pesch et al. 2011).
Subjects and methods Study population Between September 2003 and June 2010, we enrolled 1,609 men out of 2,214 active or retired workers (response rate 72.7 %) from two chemical companies. The participants were formerly exposed to carcinogenic aromatic amines and received annual invitations for a voluntary screening program. The workflow is shown in Fig. 1. Interviews, urine sampling, and processing were performed in the medical departments of the companies. Cytology and the determination of NMP22 and UroVysion were performed in a central laboratory. The ethics committee at the University of Tu¨bingen approved the study (No. 1/2003V). All participants gave written informed consent. Determination of tumor markers Tumor tests were performed according to the manufacturers (Banek et al. 2012; Huber et al. 2012). In brief, the supernatant was stabilized using the Matritech NMP22 Urine Collection KitÒ (Matritech, Freiburg, Germany) and stored at -20 °C until shipping at 4–8 °C for NMP22 determination with the Immulite 2000Ò assay, which was used until 2006, and subsequently replaced with a new NMP22 ELISAÒ (Matritech, Newton, MA). Two monoclonal antibodies (Mab302-18, Ab302-22) were employed, using 10 U/mL as cutoff for positivity. The cell pellets were stored at 4–8 °C until overnight shipment to perform
Int Arch Occup Environ Health Fig. 1 Flow diagram of the prospective UroScreen study
cytology and UroVysion. Copy number variations (CNVs) in about 25 suspicious urothelial cells were assessed with FISH at chromosomes 3, 7, 17, and 9p21 using the UroVysionTM Bladder Cancer Kit (Abbott Laboratories, Abbott Park, IL). Tests were positive if C4 nuclei showed DNA gain for C2 probes (chromosomes 3, 7, 17) or if C12 nuclei showed a loss of both alleles at 9p21. Urine status was available for 84 % of the samples. Creatinine was determined with CREA plusÒ and the cell content with Combur10 TestÒ strips (Roche Diagnostics, Mannheim, Germany). The sediment was investigated in C15 high-power fields. Leukocytes were categorized as absent, traces, or present, and hematuria was classified into microhematuria and gross hematuria (Pesch et al. 2011).
diagnosis or the last screen if non-diseased (Behrens et al. 2013). Multivariate generalized estimation equation models were applied to estimate odds ratios (OR) with 95 % confidence intervals (CI) for factors influencing the positivity of tumor tests in repetitive samples from non-diseased men with information on urine status. Analyses were performed with SAS/STAT and SAS/IML, version 9.2 (SAS Institute Inc., Cary, NC, USA).
Diagnostic follow-up
A total of 1,609 men provided 7,091 samples for urinalysis between 2003 and 2010. Non-diseased participants (n = 1,589) underwent on average five screens, with a preference for joining every other annual examination (see Table 1). A low creatinine concentration (\0.5 g/L) was observed in 16.4 % of these samples. Traces of leukocytes and microhematuria were detected in 62.3 and 18.7 % of the samples, respectively. Four out of 21 samples from cases presented with gross hematuria and three with microhematuria before diagnosis. Urinary leukocytes were frequently found in these prediagnostic samples, except in two samples with diluted urines. Compliance for cystoscopy following 341 positive tests was 66.3 % (see Fig. 1). The time between urinalysis and diagnosis of BC varied between 0 and 26 months. We observed five low-grade lesions, 13 high-grade BC, and
Positivity of cytology, NMP22, or UroVysion resulted in written recommendations for cystoscopy that were sent from the company physicians to the participants with information for an urologist of their choice. The company physicians conducted the follow-up on BC. Completeness was supported by the fact that participants with BC may have an occupational disease and are eligible for compensation. A reference pathologist classified lesions according to WHO (Sauter et al. 2004). Statistical analysis Performance measures were calculated according to accepted definitions from the test in the last sample before
Results Screening characteristics of the study population and bladder cancer cases
123
Int Arch Occup Environ Health Table 1 Screening characteristics of the UroScreen participants (1,609 retired or active chemical workers) Casesa (n = 20)
Non-cases (n = 1,589)
Age, years [median (range)]
67 (38–78) at diagnosis
62 (27–90) in 2010
Number of screens [median (range)]
2 (1–10) until diagnosis
5 (1–11)
Months between last screen and diagnosis [median (range)]
3 (0–26) Sample before diagnosis
All samples with urine statusb
6/17 (35.3 %)
964/5,895 (16.4 %)
Traces
11/15 (73.3 %)
3,456/5,549 (62.3 %)
Present
2/15 (13.3 %)
347/5,549 (6.3 %)
Samples with creatinine \0.5 mg/L [n (%)] Samples with leukocytes [n (%)]
Samples with hematuria [n (%)] Microhematuria
3/15 (20.0 %)
1,036/5,549 (18.7 %)
Gross hematuria
4/15 (26.7 %)
119/5,549 (2.1 %)
a
One participant with an incident and recurrent bladder tumor during 2003–2010
b
Urine status available for 1,316 participants
three papillomas. Age at diagnosis ranged between 38 and 78 years. Although most of the cases were diagnosed with favorable stages, we observed that a negative cystoscopy finding following a positive marker test can impair the compliance for a subsequent diagnostic workup. Performance of tumor tests Whereas NMP22 could be determined in most supernatants, the sediment was a limiting factor for cell-based assays (see Fig. 1). A total of 341 samples were tested positive for cytology (41/6,567, 0.6 %), UroVysion (107/ 6,525, 1.6 %), and/or NMP22 (224/7,058; 3.2 %). Figure 2 presents false-positive results, predominantly from the NMP22 assay. Only four out of 217 false-positive NMP22 tests shared positivity of cytology or UroVysion. Sixteen out of 30 false-positive cytology results shared a falsepositive UroVysion test. Table 2 depicts the performance measures of the tumor tests. Fifteen out of 21 lesions were detected by urine tests in the last sample before diagnosis (see Fig. 2, true-positive tests are shown in brackets). One tumor was positive in all assays. NMP22 in combination with UroVysion detected 14 tumors. Cytology was positive in eight samples, overlapping with seven of the nine true-positive UroVysion tests. UroVysion in combination with cytology would have detected 10 cases. NMP22 identified five additional lesions. Lowering the cutoff for NMP22 to 6 U/mL would have led to the detection of two additional cases in diluted urines, but at the expense of four times more cystoscopy recommendations in all participants. All tests had a high specificity, with cytology being the most specific method (see Table 2). Sensitivity was similar for cytology (44.4 %) and UroVysion (45.0 %) but low for
123
Fig. 2 False-positive tests of cytology, UroVysion, and NMP22 in 318 out of 7,000 samples from 1,589 non-diseased participants of the UroScreen study. True-positive tests are shown in brackets
NMP22 (28.6 %). Five cases presented with diluted urine samples (creatinine \0.5 g/L) and were tested negative for NMP22. NMP22 in combination with UroVysion achieved a sensitivity of 66.7 %, a specificity of 94.5 %, and a positive predictive value (PPV) of 13.7 %. The sensitivity was higher in cases attending screens within 12 months before diagnosis and when excluding papillomas (58.3 % for cytology and 85.7 % for NMP22/UroVysion). Three out of the six interval cancers had their last screen more than 1 year before diagnosis.
Int Arch Occup Environ Health Table 2 Performance measures of tumor tests based on results from cases (last sample before diagnosis) and non-cases (last sample available) Measure
Cytology
UroVysion
NMP22
NMP22/UroVysion
Cytology/UroVysion
Cytology/NMP22
All prediagnostic samples True positive
8
9
6
14
10
13
False positive
22
46
43
88
57
65
True negative
1,478
1,472
1,540
1,501
1,492
1,524
False negative
10
11
15
7
10
8
Sensitivity (%)
44.4
45.0
28.6
66.7
50.0
61.9
98.5
97.0
97.3
94.5
96.3
95.9
Positive predictive value (%)
26.7
16.4
12.2
13.7
14.9
16.7
Negative predictive value (%)
99.3
99.3
99.0
99.5
99.3
99.5
True positive False negative
0 4
0 4
1 3
1 3
0 4
1 3
Sensitivity (%)
0
0
25.0
25.0
0
25.0
Specificity (%)
[12 Months before diagnosis
B12 Months before diagnosis True positive
8
9
5
13
10
12
False negative
6
7
12
4
6
5
Sensitivity (%)
57.1
56.3
29.4
76.5
62.5
70.6
B12 Months before diagnosis and without papillomas True positive
7
9
4
12
9
10
False negative
5
5
10
2
5
4
Sensitivity (%)
58.3
64.3
28.6
85.7
64.3
71.4
Factors influencing the tumor tests Table 3 presents ORs with 95 % CIs for potential predictors of false-positive tests in more than 5,000 samples from non-diseased participants. NMP22 was more frequently positive in elderly men (OR 1.67, 95 % CI 1.11–2.51), in samples with leukocytes (OR 4.13, 95 % CI 2.60–6.56) or hematuria (OR 1.48, 95 % CI 1.01–2.16). A low creatinine concentration as proxy of a diluted urine samples was associated with lower NMP levels and hence less positive tests (OR 2.05, 95 % CI 1.07–3.94 for C0.5 vs. \0.5 g/L). In contrast, the UroVysion test was more frequently positive in diluted urines (OR 0.35, 95 % CI 0.21–0.58 for C0.5 vs. \0.5 g/L). Summary evaluation of the tumor tests Table 4 gives an overview on costs of the tumor tests in UroScreen. Calculated from what is currently reimbursed in Germany (cytology: 25 €, UroVysion: 170 €, NMP22: 29 €), the assay-specific costs are approximately 1.5 million € for all urine samples and 100,000 € per case (15 detected with cytology, UroVysion, or NMP22). UroVysion detected two cases in addition to cytology at the expense of 554,625 € per case, and NMP22 found five additional cases at assay costs of 40,936 € per case.
Cytology is highly specific but has a limited sensitivity. The UroVysion assay is expensive and shows an apparent overlap with the cytology findings. FISH requires a qualified laboratory and statistical expertise to better explore the complex data. Although NMP22 detected five additional cases, the high number of false-positive tests observed in UroScreen raises concern about using this marker in asymptomatic subjects.
Discussion Whom to target to BC screening? The concept of early detection of BC holds great promise, particularly in at-risk groups with exposure to aromatic amines, although screening in the general population is not yet standard of care (Lotan et al. 2006). BC is one of the most expensive cancers and more common than, for example, cervix cancer (Larre et al. 2013). BC screening is in accordance with criteria issued by WHO with the exception of the yet unsolved question whom to target with regard to cost efficiency (Lotan et al. 2006). The test-specific costs were about 100,000 € to detect 15 of 21 tumors in more than 7,000 urine samples collected in UroScreen using cytology, UroVysion, or NMP22. The cost efficiency
123
Int Arch Occup Environ Health Table 3 Factors influencing test positivity in repetitive samples from non-diseased participants with available urine status data Variable
Cytology N = 5,164 N (positive)
UroVysion N = 5,107 OR (95 % CI)a
N (positive)
NMP22 N = 5,495 OR (95 % CI)a
N (positive)
OR (95 % CI)a
Age (years) \60 C60 Smoking status
2,956 (7) 2,208 (9)
Never
1,454 (2)
Ever
3,710 (14)
1 1.79 (0.66–4.89) 1 2.87 (0.66–12.5)
2,905 (42) 2,202 (27) 1,435 (15) 3,672 (54)
1 0.90 (0.52–1.55) 1 1.50 (0.80–2.83)
3,153 (59) 2,342 (84) 1,559 (39) 3,936 (104)
1 1.67 (1.11–2.51) 1 1.00 (0.62–1.59)
Leukocytes None or traces Present
4,844 (14) 320 (2)
1 1.62 (0.35–7.49)
4,796 (65) 311 (4)
1
5,154 (107)
1.20 (0.47–3.08)
341 (36)
1 4.13 (2.60–6.56)
Hematuria None or traces
4,094 (12)
Present
1,070 (4)
1 1.09 (0.32–3.75)
4,052 (58) 1,055 (11)
1 0.82 (0.42–1.60)
4,355 (96) 1,140 (47)
1 1.48 (1.01–2.16)
Creatinine (g/L)
a
\0.5
821 (1)
C0.5
4,343 (15)
1 2.68 (0.36–19.87)
799 (24) 4,308 (45)
1 0.35 (0.21–0.58)
892 (11) 4,603 (132)
1 2.05 (1.07–3.94)
Odds ratio with 95 % confidence interval
Table 4 Cost calculation and summary evaluation of BC tests in UroScreen Cytology
UroVysion
NMP22
Costs per assay (€)
25
170
29
Assays
6,567
6,525
7,058
Assay costs (€)
164,175
1,109,250
204,682
Detected cases
8
9
6
Assay costs per detected case (€)
20,522
123,250
34,114
Cases detected in addition to cytology
–
2
5
Assay costs per additional case (€)
–
554,625
40,936
Advantage
High specificity
Early detection of genomic changes
Point-of-care test available
Disadvantage
Limited sensitivity
High costs
Limited specificity
Overlapping findings with cytology
would considerably improve when markers would detect early stages of the disease and when they are applied to monitor BC patients because of the high recurrence rate. In line with recommendations of the International Labor Office to provide surveillance of the workers’ health following exposure to occupational carcinogens, UroScreen was conducted within the frame of such a program of the German Social Accident Insurance for chemical workers with former exposure to aromatic amines. A previous analysis associated with this surveillance program revealed 21 BC cases in 1,679 subjects between 1994 and 1999 in contrast to 15 incident cases observed in 1,609 men between 2003 and 2010 when excluding premalignant stages and recurrent cases (Nasterlack et al. 2001). Due to
123
the increased attained age of this cohort, the number of expected cases would be higher. Hence, the standardized incidence ratio would be lower than 8.75 as reported for the previous analysis. A detailed risk assessment will be subject to another analysis with focus on exposure to aromatic amines. This lower BC incidence may reflect the longer period since carcinogenic aromatic amines have been banned from the manufacture of dyes, i.e., the decline in cases reflects the positive effects of preventive measures. How to deal with premalignant stages? Cancer screening aims to detect favorable stages before the onset of symptoms. We observed a downstaging (83 % of
Int Arch Occup Environ Health
the cases, excluding papilloma, were pTis, pTa, and pT1) in comparison with the distribution of BC in the general population (67–75 % non-muscle invasive) (Fischer et al. 1998). However, screening may lead to over-diagnosis, i.e., the detection of lesions with low premalignant potential like papillomas, the majority of which do not become malignant. Early stages are considered a step in the time course of tumor development. One case in our study had a papilloma that was not detected with markers and developed high-grade BC a few years later, which was diagnosed following elevated NMP22. The sensitivity of the marker panel increased to 86 % when papillomas were excluded from the analysis. Cystoscopy as the diagnostic gold standard may detect the tumor later than a marker. For example, genomic changes can be detected with the UroVysion assay earlier in urothelial cells exfoliated into the urine before aberrant cells can be observed with cytology or before lesions become visible in the target organ with cystoscopy. UroVysion tested more samples positive than cytology. A follow-up of these subjects is necessary to assess the positive predictive value of chromosomal aberrations. Can UroVysion in combination with NMP22 detect more bladder tumors than cytology at reasonable costs? Bladder cancer is an excellent example for the early detection of favorable stages with the non-invasive determination of tumor markers, due to the exfoliation of cancer cells or the leakage of molecular markers from the target organ into the urine. UroScreen provided evidence that the combination of a protein- and a cell-based marker can detect more cases than conventional cytology alone, a method with limited sensitivity (Sullivan et al. 2010). Out of 21 tumors observed between 2003 and 2011, 14 were detected with NMP22 or UroVysion in comparison with eight tumors with cytology. However, the laborintensive UroVysion assay overlapped with cytology, due to the preselection of atypical cells. We would suggest refraining from this selection as genomic changes may occur earlier than morphological changes of the cells would become visible. Both tests are prone to observer bias, and the costs of the FISH assay are much higher than what is currently reimbursed in Germany for cytology (170 vs. 25 €). The detection of two additional cases with UroVysion in 6,525 samples would cost 554,625 € per case. However, the strong correlation between copy number variations at three out of the four loci may likely allow a reduction in the probes and thus the development of a cheaper assay (Bonberg et al. 2013). The quantitative NMP22 assay (29 €) detected five cases in addition to cytology (40,936 € per case), but at the expense of more false-positive results, thus lowering the panel’s specificity
to 94.2 % in comparison with 98.5 % if applying cytology alone. What is the trade-off for good performance measures? In general, higher sensitivity from a panel of diagnostic markers, when compared to individual tests, can be only achieved at the expense of specificity. The trade-off between sensitivity and specificity holds for combined as well as individual tests. For example, various cutoffs have been applied to NMP22 (Abd El Gawad et al. 2005). Using 10 U/mL as recommended by the manufacturer, 3 % of our samples tested positive and six tumors were detected. The consequence of missing a subject with invasive BC favors a test with high sensitivity. A cutoff of 6 U/mL would have identified two more cases with the trade-off of four times more false-positive tests. Also, the biological features of the markers may influence their performance for BC screening. Particularly, protein-based markers may have a limited specificity as they can be involved in a variety of diseases in addition to cancer such as urinary tract infections. Nuclear matrix proteins are associated with DNA replication and RNA synthesis (Kumara-Siri et al. 1986). UroVysion seems to be more specific as it detects cancer-related genomic changes. For example, DNA loss at 9p21 has been frequently observed in cancer cells, which hosts the tumor suppressor gene p16CDKNA2 (Aravidis et al. 2012; Klorin et al. 2013; Panani et al. 2009). Diagnostic markers can only be applied in asymptomatic subjects at an acceptable rate of false-positive findings, usually not exceeding 2 % due to the invasiveness of the diagnostic workup and psychological distress (Pepe et al. 2008). Only cytology was associated with such an excellent rate but suffered from a limited sensitivity. Higher rates may be acceptable for the detection of aggressive forms like lung cancer or when monitoring patients for recurrent BC where cystoscopy is part of the medical follow-up program. Because marker validation integrated into prospective designs or surveillance programs for occupationally exposed workers is yet a bottleneck in research, performance measures from other screening studies have been rarely published (Brenner and Normolle 2007; Ransohoff 2013). So far, no randomized controlled trial which is the most rigorous study design for assessing the benefit of BC screening with markers has been performed (Larre et al. 2013). By contrast, imaging methods for the early detection of cancer were comprehensively evaluated with the National Lung Cancer Screening Trial as prominent example. Sensitivity and specificity varied by method (low-dose CT: 94.4 and 72.6 %, radiography: 59.6 and 94.1 %, respectively) (Aberle et al. 2013). Similar performance measures may likely occur in occupational settings,
123
Int Arch Occup Environ Health
for example, in lung cancer screening of asbestos-exposed workers. Voluntary screening programs with asymptomatic subjects may yield worse measures of the performance of markers than reported from clinical settings with symptomatic subjects or patients because of the smaller tumor volume at early stages and a lower compliance of healthy volunteers to attend regular screens. In addition, we observed a low compliance to repeat screens and a limited compliance for cystoscopy following a positive finding. Despite a good overall response rate in UroScreen, few men took part on an annual basis. Three out of the six interval cases had their last screen more than 1 year before diagnosis. Cytology and UroVysion were negative in all samples collected more than 12 months before diagnosis. The sensitivity of the marker panel increased to 76.5 % when restricted to prediagnostic samples from cases that underwent screens within 1 year of diagnosis. This observation is in line with a validation study on the performance of CA125 to detect ovarian cancer (Zhu et al. 2011). Does the urine status influence the test results with molecular markers? Currently, there is no consensus about the consideration of the urine status with regard to the test results. This may be due to a lack of detailed data from sufficiently large studies. Based on more than 5,000 urine samples from nondiseased participants, we confirmed that leukocytes are associated with false-positive NMP22 tests (Lotan and Shariat 2008; Sharma et al. 1999). Asymptomatic urinary tract infection can be a concomitant condition in the elderly, and a single positive NMP22 test should not initiate an invasive diagnostic procedure if supportive diagnostic data fail. The test could be repeated but we observed a low compliance when a participant was invited for a second examination (Pesch et al. 2011). On the other hand, false-negative NMP22 tests can occur in diluted urines. Frequently, urine was voided shortly before an examination although the participants were informed about the collection of spot urines. In diluted urines, NMP22 may not have reached the cutoff—as observed in two of the nondetected cases. Creatinine adjustment could be applied but may be less feasible in clinical practice than in biomonitoring studies (Todenhoefer et al. 2012). The cell content of the urine sample was a limiting factor for the cell-based tests. Samples with low creatinine content were frequently not suitable for UroVysion. However, more cells with genomic changes were detected in freshly voided urines but rarely in urines with unusually high creatinine. To address this more than tenfold variation in test positivity, we suggest collecting two urine samples, one as spot urine and a second sample after physical activity.
123
Who should receive a recommendation for a diagnostic follow-up? When applying marker panels, the rate of false-positive findings usually increases and a decision is needed to better classify these subjects into those who should repeat the tests and those who should undergo an invasive diagnostic workup. However, a consensus on cystoscopy is still lacking (Lotan et al. 2009a). It also depends on the target population whether asymptomatic subjects or BC patients are subject to screening. False-positive tests may be of less concern in BC patients where regular cystoscopy is part of the surveillance guidelines (Babjuk et al. 2011). Screening of asymptomatic volunteers requires more restrictive rules for the initiation of an invasive diagnostic workup. Subjects with abundant leukocytes should be carefully assessed before cystoscopy is recommended. For example, NMP22 concentration was highest in a dialysis patient. It is important to note that a third of the volunteers with a positive test result refused to undergo cystoscopy. We also observed some resistance to a subsequent invasive diagnostic workup when the screenee was repeatedly tested positive. This may be of relevance for subjects with urinary tract infection or inconclusive cytology. A limited specificity of diagnostic markers may require additional steps prior to an invasive diagnostic workup to minimize false-positive findings. However, complex decisions are rarely feasible in clinical routine. A recent evaluation of compliance with guidelines for BC patients revealed that only one out of 4,545 patients received all recommended measures (Chamie et al. 2011). This may imply robust tests and specialized centers with expertise in the conduct and interpretation of complex assays.
Conclusions A number of aromatic amines have been classified as human carcinogens and consequently were subjected to regulation to limit occupational exposure. In addition to efficient measures applied to improve primary prevention and in line with recommendations of the ILO, the German Social Accident Insurance offers health surveillance programs to workers with former exposure to occupational carcinogens. BC is an excellent screening target for urinary markers. Marker performance improved considerably when participants complied with the screening interval of 1 year. UroScreen revealed that protein- and cell-based markers can complement each other in the early detection of BC. However, NMP22 was associated with many false-positive tests, in particular in the presence of leukocytes. As long as UroVysion remains costly and preselects atypical cells, it will not be a useful adjunct to cytology. The relatively low
Int Arch Occup Environ Health
incidence of BC is a limiting factor when balancing benefits and cost efficiency. However, markers may be valuable for monitoring patients with BC due to the high recurrence rate of this cancer. Acknowledgments We thank Friedhelm Eberle for the careful execution and documentation of the urinalyses. This unique database allowed the exploration of the influence of the urine status on the tumor tests. Conflict of interest The study received a Grant (FB093) from the German Social Accident Insurance (DGUV), Sankt Augustin, Germany. A.S. received free assay kits from Abbott Laboratories and Matritech/Alere Ltd. G.J. received free assay reagents from Fujirebio Diagnostics, Inc. M.K. and H.W. are employees of the sponsor. Y.L. does research with Abbott Laboratories and Matritech/Alere Ltd.
Appendix UroScreen Study Group BASF SE, Department of Occupational Medicine and Health Protection, Ludwigshafen: Bernd Scheuermann, Friedhelm Eberle, Thomas Mayer, Michael Nasterlack. BG RCI, Bereich Pra¨vention, Heidelberg: Harald Wellha¨ußer, Matthias Kluckert; Organisationsdienst fu¨r nachgehende Untersuchungen (ODIN): Reinhard Detzner. Institute for Prevention and Occupational Medicine of the German Social Accident Insurance, Institute of RuhrUniversita¨t Bochum (IPA), Bochum: Dirk Taeger, Beate Pesch, Katarzyna Gawrych, Heike Bontrup, Georg Johnen, Nadine Bonberg, Judith Delbanco, Evelyn Heinze, Thomas Bru¨ning. Protein Research Unit Ruhr within Europe (PURE), Ruhr University of Bochum: Nadine Bonberg. Currenta GmbH & Co. OHG, Safety—Health Protection, Leverkusen: Gabriele Leng. Institute of Urology, Eberhard Karls University, Tu¨bingen: Gerhard Feil, Karl-Dietrich Sievert, Se´verine Banek, Christian Schwentner, Margarete Geiger, Erika Senger, Valentina Gerber, Andrea Hohneder, Gundi Beger, Ursula Kuehs, Jo¨rg Hennenlotter, Arnulf Stenzl.
References Abd El Gawad I, Moussa HS, Nasr MI, El Gemae EH, Masooud AM, Ibrahim IK, El Hifnawy NM (2005) Comparative study of NMP22, telomerase, and BTA in the detection of bladder cancer. J Egypt Natl Canc Inst 17:193–202 Aberle DR, DeMello S, Berg CD, Black WC, Brewer B, Church TR, Clingan KL, Duan F, Fagerstrom RM, Gareen IF et al (2013) Results of the two incidence screenings in the national lung screening trial. N Engl J Med 369:920–931
Aravidis C, Panani AD, Kosmaidou Z, Thomakos N, Rodolakis A, Antsaklis A (2012) Detection of numerical abnormalities of chromosome 9 and p16/CDKN2A gene alterations in ovarian cancer with fish analysis. Anticancer Res 32:5309–5313 Babjuk M, Oosterlinck W, Sylvester R, Kaasinen E, Bohle A, PalouRedorta J, Roupret M (2011) EAU guidelines on non-muscleinvasive urothelial carcinoma of the bladder, the 2011 update. Eur Urol 59:997–1008 Banek S, Schwentner C, Taeger D, Pesch B, Nasterlack M, Leng G, Gawrych K, Bonberg N, Johnen G, Kluckert M et al (2012) Prospective evaluation of fluorescence-in situ-hybridization to detect bladder cancer: results from the UroScreen-study. Urol Oncol. doi:10.1016/j.urolonc.2012.04.015 Behrens T, Bonberg N, Casjens S, Pesch B, Bru¨ning T (2013) A practical guide to epidemiological practice and standards in the identification and validation of diagnostic markers using a bladder cancer example. BBA Proteins Proteomics. doi:10.1016/ j.bbpap.2013. Article first published online: 18 July 2013 Bonberg N, Taeger D, Gawrych K, Johnen G, Banek S, Schwentner C, Sievert KD, Wellhausser H, Kluckert M, Leng G et al (2013) Chromosomal instability and bladder cancer—the UroVysionTM test in the UroScreen study. BJU Int. doi:10.1111/j.1464-410X. 2012.11666.x Brenner DE, Normolle DP (2007) Biomarkers for cancer risk, early detection, and prognosis: the validation conundrum. Cancer Epidemiol Biomarkers Prev 16:1918–1920 Chamie K, Saigal CS, Lai J, Hanley JM, Setodji CM, Konety BR, Litwin MS (2011) Compliance with guidelines for patients with bladder cancer: variation in the delivery of care. Cancer 117:5392–5401 FDA, U.S.Federal Food and Drug Administration (2000) NMP22. http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfTopic/pma/ pma.cfm?num=P940035S002. Accessed 29 May 2013 FDA, U.S.Federal Food and Drug Administration (2005) UroVysion. http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfTopic/pma/ pma.cfm?num=P030052. Accessed 29 May 2013 Fischer CG, Waechter W, Kraus S, Fuentecilla PE, Weidner W, Dudeck J (1998) Urologic tumors in the Federal Republic of Germany: data on 56,013 cases from hospital cancer registries. Cancer 82:775–783 Freudenthal RI, Stephens E, Anderson DP (1999) Determining the potential of aromatic amines to induce cancer of the urinary bladder. Int J Toxicol 18:353–359 Grossman HB, Messing E, Soloway M, Tomera K, Katz G, Berger Y, Shen Y (2005) Detection of bladder cancer using a point-of-care proteomic assay. JAMA 293:810–816 Hemstreet GP III, Yin S, Ma Z, Bonner RB, Bi W, Rao JY, Zang M, Zheng Q, Bane B, Asal N et al (2001) Biomarker risk assessment and bladder cancer detection in a cohort exposed to benzidine. J Natl Cancer Inst 93:427–436 Huber S, Schwentner C, Taeger D, Pesch B, Nasterlack M, Leng G, Mayer T, Gawrych K, Bonberg N, Pelster M et al (2012) Nuclear matrix protein-22: a prospective evaluation in a population at risk for bladder cancer. Results from the UroScreen study. BJU Int 110:699–708 Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D (2011) Global cancer statistics. CA Cancer J Clin 61:69–90 Johnen G, Gawrych K, Bontrup H, Pesch B, Taeger D, Banek S, Kluckert M, Wellhausser H, Eberle F, Nasterlack M et al (2012) Performance of survivin mRNA as a biomarker for bladder cancer in the prospective study UroScreen. PLoS One 7:e35363 Klorin G, Rozenblum E, Glebov O, Walker RL, Park Y, Meltzer PS, Kirsch IR, Kaye FJ, Roschke AV (2013) Integrated highresolution array CGH and SKY analysis of homozygous deletions and other genomic alterations present in malignant mesothelioma cell lines. Cancer Genet 206:191–205
123
Int Arch Occup Environ Health Kumara-Siri MH, Shapiro LE, Surks MI (1986) Association of the 3,5,3’-triiodo-L-thyronine nuclear receptor with the nuclear matrix of cultured growth hormone-producing rat pituitary tumor cells (GC cells). J Biol Chem 261:2844–2852 Larre S, Catto JW, Cookson MS, Messing EM, Shariat SF, Soloway MS, Svatek RS, Lotan Y, Zlotta AR, Grossman HB (2013) Screening for bladder cancer: rationale, limitations, whom to target, and perspectives. Eur Urol 63:1049–1058 Lokeshwar VB, Schroeder GL, Selzer MG, Hautmann SH, Posey JT, Duncan RC, Watson R, Rose L, Markowitz S, Soloway MS (2002) Bladder tumor markers for monitoring recurrence and screening comparison of hyaluronic acid-hyaluronidase and BTA-Stat tests. Cancer 95:61–72 Lokeshwar VB, Habuchi T, Grossman HB, Murphy WM, Hautmann SH, Hemstreet GP III, Bono AV, Getzenberg RH, Goebell P, Schmitz-Drager BJ et al (2005) Bladder tumor markers beyond cytology: international consensus panel on bladder tumor markers. Urology 66:35–63 Lotan Y, Shariat SF (2008) Impact of risk factors on the performance of the nuclear matrix protein 22 point-of-care test for bladder cancer detection. BJU Int 101:1362–1367 Lotan Y, Svatek RS, Sagalowsky AI (2006) Should we screen for bladder cancer in a high-risk population? A cost per life-year saved analysis. Cancer 107:982–990 Lotan Y, Capitanio U, Shariat SF, Hutterer GC, Karakiewicz PI (2009a) Impact of clinical factors, including a point-of-care nuclear matrix protein-22 assay and cytology, on bladder cancer detection. BJU Int 103:1368–1374 Lotan Y, Elias K, Svatek RS, Bagrodia A, Nuss G, Moran B, Sagalowsky AI (2009b) Bladder cancer screening in a high risk asymptomatic population using a point of care urine based protein tumor marker. J Urol 182:52–57 Nasterlack M, Scheuermann B, Messerer P, Pallapies D, Zober A (2001) Harnblasenkrebs in einem Risikokollektiv: Klinische und epidemiologische Aspekte. Symp Med 12:17–19 Panani AD, Maliaga K, Babanaraki A, Bellenis I (2009) Numerical abnormalities of chromosome 9 and p16CDKN2A gene deletion detected by FISH in non-small cell lung cancer. Anticancer Res 29:4483–4487 Pepe MS, Feng Z, Janes H, Bossuyt PM, Potter JD (2008) Pivotal evaluation of the accuracy of a biomarker used for classification
123
or prediction: standards for study design. J Natl Cancer Inst 100:1432–1438 Pesch B, Nasterlack M, Eberle F, Bonberg N, Taeger D, Leng G, Feil G, Johnen G, Ickstadt K, Kluckert M et al (2011) The role of haematuria in bladder cancer screening among men with former occupational exposure to aromatic amines. BJU Int 108:546–552 Ransohoff DF (2013) Cultivating cohort studies for observational translational research. Cancer Epidemiol Biomarkers Prev 22:481–484 Rehn L (1895) Blasengeschwu¨lste bei Fuchsin-Arbeitern. Archiv fu¨r Klinische Chirurgie 50:588–600 Sauter G, Algaba F, Amin M, Busch C, Cheville J, Gasser T, Grignon D, Hofstaedter F, Lopez-Beltran A, Epstein J (2004) Noninvasive urothelial neoplasias: WHO classification of noninvasive papillary urothelial tumors. In: Eble JN, Sauter G, Epstein JI, Sesterhenn IA (eds) World Health Organization classification of tumors. Pathology and genetics of tumors of the urinary system and male genital organs, 3rd edn. IARC Press, Lyon Shariat SF, Karam JA, Lotan Y, Karakiewizc PI (2008) Critical evaluation of urinary markers for bladder cancer detection and monitoring. Rev Urol 10:120–135 Sharma S, Zippe CD, Pandrangi L, Nelson D, Agarwal A (1999) Exclusion criteria enhance the specificity and positive predictive value of NMP22 and BTA stat. J Urol 162:53–57 Sullivan P, Chan J, Levin M, Rao J (2010) Urine cytology and adjunct markers for detection and surveillance of bladder cancer. Am J Transl Res 2:412–440 Todenhoefer T, Hennenlotter J, Witstruk M, Gakis G, Aufderklamm S, Kuehs U, Stenzl A, Schwentner C (2012) Influence of renal excretory function on the performance of urine based markers to detect bladder cancer. J Urol 187:68–73 van Rhijn BW, van der Poel HG, Van der Kwast TH (2005) Urine markers for bladder cancer surveillance: a systematic review. Eur Urol 47:736–748 Vineis P, Pirastu R (1997) Aromatic amines and cancer. Cancer Causes Control 8:346–355 Zhu CS, Pinsky PF, Cramer DW, Ransohoff DF, Hartge P, Pfeiffer RM, Urban N, Mor G, Bast RC Jr, Moore LE et al (2011) A framework for evaluating biomarkers for early detection: validation of biomarker panels for ovarian cancer. Cancer Prev Res (Phila) 4:375–383
ORIGINAL ARTICLE
Screening for bladder cancer with urinary tumor markers in chemical workers with exposure to aromatic amines Beate Pesch • Dirk Taeger • Georg Johnen • Katarzyna Gawrych • Nadine Bonberg • Christian Schwentner • Harald Wellha¨ußer • Matthias Kluckert • Gabriele Leng • Michael Nasterlack • Yair Lotan • Arnulf Stenzl • Thomas Bru¨ning • the UroScreen Study Group
Received: 31 May 2013 / Accepted: 2 October 2013 Ó Springer-Verlag Berlin Heidelberg 2013
Abstract Purpose To validate urinary markers for the early detection of bladder cancer (BC) in chemical workers. Methods UroScreen was conducted as a validation study for tumor markers within the frame of a health surveillance program of the German Social Accident Insurance for active or retired workers with former exposure to aromatic amines. From 2003 to 2010, 1,609 men took part in voluntary annual screens. Cytology, the quantitative NMP22Ò assay, and UroVysionTM were applied to 7,091 urine samples. Results Fifteen out of 21 tumors were detected following test positivity. The UroVysion/NMP22 panel detected 14 out of 21 tumors versus 8 tumors with cytology alone Beate Pesch and Dirk Taeger have equally contributed to this article. Please refer the ‘‘Appendix’’ section for UroScreen Study Group.
(sensitivity 66.7 vs. 44.4 %, specificity 94.5 vs. 98.5 %). The sensitivity of the panel increased to 85.7 % in samples collected B12 months before diagnosis and when papillomas were excluded, compared to 58.3 % with cytology. About 3 % of NMP22 tests were false-positive. UroVysion results overlapped with cytology due to the preselection of atypical cells. NMP22 was less and UroVysion more frequently positive in diluted urine samples. Leukocytes confounded NMP22 but not UroVysion. The low incidence of BC in this study population yielded low positive predictive values of the markers and high costs per tumor detected with screening. Conclusions UroVysion in combination with NMP22 detected more cases than cytology alone, at the expense of a lower specificity. High costs per detected case resulted from a lower BC incidence than in the past when levels of occupational exposure to aromatic amines were higher. Currently, it cannot be recommended to apply these
B. Pesch (&) D. Taeger G. Johnen K. Gawrych N. Bonberg M. Kluckert T. Bru¨ning Institute for Prevention and Occupational Medicine of the German Social Accident Insurance, Institute of the RuhrUniversita¨t Bochum (IPA), Buerkle-de-la-Camp-Platz 1, 44789 Bochum, Germany e-mail: [email protected]
G. Leng Department of Health Protection and Occupational Safety, Currenta GmbH & Co. OHG, Leverkusen, Germany
N. Bonberg Protein Research Unit Ruhr within Europe (PURE), Ruhr University of Bochum, Bochum, Germany
Y. Lotan Department of Urology, University of Texas Southwestern Medical Center, Dallas, TX, USA
M. Nasterlack Occupational Medicine and Health Protection, BASF SE, Ludwigshafen, Germany
C. Schwentner A. Stenzl Institute of Urology, Eberhard Karls University, Tu¨bingen, Germany H. Wellha¨ußer M. Kluckert German Social Accident Insurance’s Institution for the Raw Materials and Chemical Industry (BG RCI), Heidelberg, Germany
123
Int Arch Occup Environ Health
markers for screening in asymptomatic workers. The increase in sensitivity is not balanced by the high costs of UroVysion and the false-positive tests of NMP22. Keywords Aromatic amines Bladder cancer Chemical workers Cytology NMP22 Screening UroVysion Validation study
Introduction Every year, bladder cancer (BC) accounts for approximately 400,000 new cases and 150,000 deaths worldwide (Jemal et al. 2011). Occupational exposure to aromatic amines has been linked to bladder cancer (Vineis and Pirastu 1997). The recognition of occupational risks dates back to the observation of bladder cancer among German aniline workers (Rehn 1895) and was attributed to exposure to 4-aminobiphenyl and 2-naphthylamine (Freudenthal et al. 1999). Both arylamines, as well as benzidine and ortho-toluidine, have been classified as human carcinogens (http://monographs.iarc.fr/). Aromatic amines are also constituents of tobacco smoke and emissions from hot tar applications or coke production, which are known to entail an increased bladder cancer risk. The International Labour Organization (ILO) has recommended health surveillance for workers with exposure to occupational carcinogens. The German Social Accident Insurance (DGUV) has established a surveillance program for workers with former exposure to carcinogenic aromatic amines aimed at the early detection of BC (Nasterlack et al. 2001). The early detection of BC with non-invasive tumor markers holds, in principle, great promise (Larre et al. 2013). Most patients with BC are diagnosed after blood is detected in the urine. Urethrocystoscopy, aided by voided urine cytology, is the gold standard for BC diagnosis. However, this invasive method is not applicable to screen asymptomatic subjects, and cytology alone suffers from a limited sensitivity (Shariat et al. 2008). Urinary markers may support the non-invasive detection of BC at early stages (Lotan et al. 2006; van Rhijn et al. 2005). Assays for the determination of nuclear matrix protein 22 (NMP22) and chromosomal instability with fluorescence in situ hybridization (FISH) have been approved for the early detection of BC (FDA 2000, 2005). BladderChek as a point-of-care test of elevated NMP22 levels has been already introduced into clinical practice. A current review revealed that no study assessed the diagnostic performance of urinary markers in the context of screening (Larre et al. 2013). The lack of a randomized controlled trial (RCT) and the shortage of prospective validation studies render performance measures for already approved markers like NMP22 inconclusive (Lokeshwar
123
et al. 2005). Factors influencing the performance of tests have not yet been sufficiently assessed. Several screening studies have been performed to detect recurrent BC in patients; but few asymptomatic at-risk cohorts have been subjected to screening, like workers exposed to aromatic amines (Hemstreet et al. 2001; Lokeshwar et al. 2002). Other studies conducted with at-risk populations were performed in clinical settings but without repeated screens (Grossman et al. 2005; Lotan et al. 2009b). UroScreen took advantage of the German surveillance program for chemical workers and offered tumor tests with urinary markers in addition to cytological examinations and to the detection of hematuria in annual screens. This prospective validation study aimed to evaluate whether a panel of tumor markers detects more primary BC cases than cytology and which factors influence the performance measures. Here, we report on the performance of NMP22 in combination with UroVysion. Detailed reports have been published elsewhere (Banek et al. 2012; Bonberg et al. 2013; Huber et al. 2012; Johnen et al. 2012; Pesch et al. 2011).
Subjects and methods Study population Between September 2003 and June 2010, we enrolled 1,609 men out of 2,214 active or retired workers (response rate 72.7 %) from two chemical companies. The participants were formerly exposed to carcinogenic aromatic amines and received annual invitations for a voluntary screening program. The workflow is shown in Fig. 1. Interviews, urine sampling, and processing were performed in the medical departments of the companies. Cytology and the determination of NMP22 and UroVysion were performed in a central laboratory. The ethics committee at the University of Tu¨bingen approved the study (No. 1/2003V). All participants gave written informed consent. Determination of tumor markers Tumor tests were performed according to the manufacturers (Banek et al. 2012; Huber et al. 2012). In brief, the supernatant was stabilized using the Matritech NMP22 Urine Collection KitÒ (Matritech, Freiburg, Germany) and stored at -20 °C until shipping at 4–8 °C for NMP22 determination with the Immulite 2000Ò assay, which was used until 2006, and subsequently replaced with a new NMP22 ELISAÒ (Matritech, Newton, MA). Two monoclonal antibodies (Mab302-18, Ab302-22) were employed, using 10 U/mL as cutoff for positivity. The cell pellets were stored at 4–8 °C until overnight shipment to perform
Int Arch Occup Environ Health Fig. 1 Flow diagram of the prospective UroScreen study
cytology and UroVysion. Copy number variations (CNVs) in about 25 suspicious urothelial cells were assessed with FISH at chromosomes 3, 7, 17, and 9p21 using the UroVysionTM Bladder Cancer Kit (Abbott Laboratories, Abbott Park, IL). Tests were positive if C4 nuclei showed DNA gain for C2 probes (chromosomes 3, 7, 17) or if C12 nuclei showed a loss of both alleles at 9p21. Urine status was available for 84 % of the samples. Creatinine was determined with CREA plusÒ and the cell content with Combur10 TestÒ strips (Roche Diagnostics, Mannheim, Germany). The sediment was investigated in C15 high-power fields. Leukocytes were categorized as absent, traces, or present, and hematuria was classified into microhematuria and gross hematuria (Pesch et al. 2011).
diagnosis or the last screen if non-diseased (Behrens et al. 2013). Multivariate generalized estimation equation models were applied to estimate odds ratios (OR) with 95 % confidence intervals (CI) for factors influencing the positivity of tumor tests in repetitive samples from non-diseased men with information on urine status. Analyses were performed with SAS/STAT and SAS/IML, version 9.2 (SAS Institute Inc., Cary, NC, USA).
Diagnostic follow-up
A total of 1,609 men provided 7,091 samples for urinalysis between 2003 and 2010. Non-diseased participants (n = 1,589) underwent on average five screens, with a preference for joining every other annual examination (see Table 1). A low creatinine concentration (\0.5 g/L) was observed in 16.4 % of these samples. Traces of leukocytes and microhematuria were detected in 62.3 and 18.7 % of the samples, respectively. Four out of 21 samples from cases presented with gross hematuria and three with microhematuria before diagnosis. Urinary leukocytes were frequently found in these prediagnostic samples, except in two samples with diluted urines. Compliance for cystoscopy following 341 positive tests was 66.3 % (see Fig. 1). The time between urinalysis and diagnosis of BC varied between 0 and 26 months. We observed five low-grade lesions, 13 high-grade BC, and
Positivity of cytology, NMP22, or UroVysion resulted in written recommendations for cystoscopy that were sent from the company physicians to the participants with information for an urologist of their choice. The company physicians conducted the follow-up on BC. Completeness was supported by the fact that participants with BC may have an occupational disease and are eligible for compensation. A reference pathologist classified lesions according to WHO (Sauter et al. 2004). Statistical analysis Performance measures were calculated according to accepted definitions from the test in the last sample before
Results Screening characteristics of the study population and bladder cancer cases
123
Int Arch Occup Environ Health Table 1 Screening characteristics of the UroScreen participants (1,609 retired or active chemical workers) Casesa (n = 20)
Non-cases (n = 1,589)
Age, years [median (range)]
67 (38–78) at diagnosis
62 (27–90) in 2010
Number of screens [median (range)]
2 (1–10) until diagnosis
5 (1–11)
Months between last screen and diagnosis [median (range)]
3 (0–26) Sample before diagnosis
All samples with urine statusb
6/17 (35.3 %)
964/5,895 (16.4 %)
Traces
11/15 (73.3 %)
3,456/5,549 (62.3 %)
Present
2/15 (13.3 %)
347/5,549 (6.3 %)
Samples with creatinine \0.5 mg/L [n (%)] Samples with leukocytes [n (%)]
Samples with hematuria [n (%)] Microhematuria
3/15 (20.0 %)
1,036/5,549 (18.7 %)
Gross hematuria
4/15 (26.7 %)
119/5,549 (2.1 %)
a
One participant with an incident and recurrent bladder tumor during 2003–2010
b
Urine status available for 1,316 participants
three papillomas. Age at diagnosis ranged between 38 and 78 years. Although most of the cases were diagnosed with favorable stages, we observed that a negative cystoscopy finding following a positive marker test can impair the compliance for a subsequent diagnostic workup. Performance of tumor tests Whereas NMP22 could be determined in most supernatants, the sediment was a limiting factor for cell-based assays (see Fig. 1). A total of 341 samples were tested positive for cytology (41/6,567, 0.6 %), UroVysion (107/ 6,525, 1.6 %), and/or NMP22 (224/7,058; 3.2 %). Figure 2 presents false-positive results, predominantly from the NMP22 assay. Only four out of 217 false-positive NMP22 tests shared positivity of cytology or UroVysion. Sixteen out of 30 false-positive cytology results shared a falsepositive UroVysion test. Table 2 depicts the performance measures of the tumor tests. Fifteen out of 21 lesions were detected by urine tests in the last sample before diagnosis (see Fig. 2, true-positive tests are shown in brackets). One tumor was positive in all assays. NMP22 in combination with UroVysion detected 14 tumors. Cytology was positive in eight samples, overlapping with seven of the nine true-positive UroVysion tests. UroVysion in combination with cytology would have detected 10 cases. NMP22 identified five additional lesions. Lowering the cutoff for NMP22 to 6 U/mL would have led to the detection of two additional cases in diluted urines, but at the expense of four times more cystoscopy recommendations in all participants. All tests had a high specificity, with cytology being the most specific method (see Table 2). Sensitivity was similar for cytology (44.4 %) and UroVysion (45.0 %) but low for
123
Fig. 2 False-positive tests of cytology, UroVysion, and NMP22 in 318 out of 7,000 samples from 1,589 non-diseased participants of the UroScreen study. True-positive tests are shown in brackets
NMP22 (28.6 %). Five cases presented with diluted urine samples (creatinine \0.5 g/L) and were tested negative for NMP22. NMP22 in combination with UroVysion achieved a sensitivity of 66.7 %, a specificity of 94.5 %, and a positive predictive value (PPV) of 13.7 %. The sensitivity was higher in cases attending screens within 12 months before diagnosis and when excluding papillomas (58.3 % for cytology and 85.7 % for NMP22/UroVysion). Three out of the six interval cancers had their last screen more than 1 year before diagnosis.
Int Arch Occup Environ Health Table 2 Performance measures of tumor tests based on results from cases (last sample before diagnosis) and non-cases (last sample available) Measure
Cytology
UroVysion
NMP22
NMP22/UroVysion
Cytology/UroVysion
Cytology/NMP22
All prediagnostic samples True positive
8
9
6
14
10
13
False positive
22
46
43
88
57
65
True negative
1,478
1,472
1,540
1,501
1,492
1,524
False negative
10
11
15
7
10
8
Sensitivity (%)
44.4
45.0
28.6
66.7
50.0
61.9
98.5
97.0
97.3
94.5
96.3
95.9
Positive predictive value (%)
26.7
16.4
12.2
13.7
14.9
16.7
Negative predictive value (%)
99.3
99.3
99.0
99.5
99.3
99.5
True positive False negative
0 4
0 4
1 3
1 3
0 4
1 3
Sensitivity (%)
0
0
25.0
25.0
0
25.0
Specificity (%)
[12 Months before diagnosis
B12 Months before diagnosis True positive
8
9
5
13
10
12
False negative
6
7
12
4
6
5
Sensitivity (%)
57.1
56.3
29.4
76.5
62.5
70.6
B12 Months before diagnosis and without papillomas True positive
7
9
4
12
9
10
False negative
5
5
10
2
5
4
Sensitivity (%)
58.3
64.3
28.6
85.7
64.3
71.4
Factors influencing the tumor tests Table 3 presents ORs with 95 % CIs for potential predictors of false-positive tests in more than 5,000 samples from non-diseased participants. NMP22 was more frequently positive in elderly men (OR 1.67, 95 % CI 1.11–2.51), in samples with leukocytes (OR 4.13, 95 % CI 2.60–6.56) or hematuria (OR 1.48, 95 % CI 1.01–2.16). A low creatinine concentration as proxy of a diluted urine samples was associated with lower NMP levels and hence less positive tests (OR 2.05, 95 % CI 1.07–3.94 for C0.5 vs. \0.5 g/L). In contrast, the UroVysion test was more frequently positive in diluted urines (OR 0.35, 95 % CI 0.21–0.58 for C0.5 vs. \0.5 g/L). Summary evaluation of the tumor tests Table 4 gives an overview on costs of the tumor tests in UroScreen. Calculated from what is currently reimbursed in Germany (cytology: 25 €, UroVysion: 170 €, NMP22: 29 €), the assay-specific costs are approximately 1.5 million € for all urine samples and 100,000 € per case (15 detected with cytology, UroVysion, or NMP22). UroVysion detected two cases in addition to cytology at the expense of 554,625 € per case, and NMP22 found five additional cases at assay costs of 40,936 € per case.
Cytology is highly specific but has a limited sensitivity. The UroVysion assay is expensive and shows an apparent overlap with the cytology findings. FISH requires a qualified laboratory and statistical expertise to better explore the complex data. Although NMP22 detected five additional cases, the high number of false-positive tests observed in UroScreen raises concern about using this marker in asymptomatic subjects.
Discussion Whom to target to BC screening? The concept of early detection of BC holds great promise, particularly in at-risk groups with exposure to aromatic amines, although screening in the general population is not yet standard of care (Lotan et al. 2006). BC is one of the most expensive cancers and more common than, for example, cervix cancer (Larre et al. 2013). BC screening is in accordance with criteria issued by WHO with the exception of the yet unsolved question whom to target with regard to cost efficiency (Lotan et al. 2006). The test-specific costs were about 100,000 € to detect 15 of 21 tumors in more than 7,000 urine samples collected in UroScreen using cytology, UroVysion, or NMP22. The cost efficiency
123
Int Arch Occup Environ Health Table 3 Factors influencing test positivity in repetitive samples from non-diseased participants with available urine status data Variable
Cytology N = 5,164 N (positive)
UroVysion N = 5,107 OR (95 % CI)a
N (positive)
NMP22 N = 5,495 OR (95 % CI)a
N (positive)
OR (95 % CI)a
Age (years) \60 C60 Smoking status
2,956 (7) 2,208 (9)
Never
1,454 (2)
Ever
3,710 (14)
1 1.79 (0.66–4.89) 1 2.87 (0.66–12.5)
2,905 (42) 2,202 (27) 1,435 (15) 3,672 (54)
1 0.90 (0.52–1.55) 1 1.50 (0.80–2.83)
3,153 (59) 2,342 (84) 1,559 (39) 3,936 (104)
1 1.67 (1.11–2.51) 1 1.00 (0.62–1.59)
Leukocytes None or traces Present
4,844 (14) 320 (2)
1 1.62 (0.35–7.49)
4,796 (65) 311 (4)
1
5,154 (107)
1.20 (0.47–3.08)
341 (36)
1 4.13 (2.60–6.56)
Hematuria None or traces
4,094 (12)
Present
1,070 (4)
1 1.09 (0.32–3.75)
4,052 (58) 1,055 (11)
1 0.82 (0.42–1.60)
4,355 (96) 1,140 (47)
1 1.48 (1.01–2.16)
Creatinine (g/L)
a
\0.5
821 (1)
C0.5
4,343 (15)
1 2.68 (0.36–19.87)
799 (24) 4,308 (45)
1 0.35 (0.21–0.58)
892 (11) 4,603 (132)
1 2.05 (1.07–3.94)
Odds ratio with 95 % confidence interval
Table 4 Cost calculation and summary evaluation of BC tests in UroScreen Cytology
UroVysion
NMP22
Costs per assay (€)
25
170
29
Assays
6,567
6,525
7,058
Assay costs (€)
164,175
1,109,250
204,682
Detected cases
8
9
6
Assay costs per detected case (€)
20,522
123,250
34,114
Cases detected in addition to cytology
–
2
5
Assay costs per additional case (€)
–
554,625
40,936
Advantage
High specificity
Early detection of genomic changes
Point-of-care test available
Disadvantage
Limited sensitivity
High costs
Limited specificity
Overlapping findings with cytology
would considerably improve when markers would detect early stages of the disease and when they are applied to monitor BC patients because of the high recurrence rate. In line with recommendations of the International Labor Office to provide surveillance of the workers’ health following exposure to occupational carcinogens, UroScreen was conducted within the frame of such a program of the German Social Accident Insurance for chemical workers with former exposure to aromatic amines. A previous analysis associated with this surveillance program revealed 21 BC cases in 1,679 subjects between 1994 and 1999 in contrast to 15 incident cases observed in 1,609 men between 2003 and 2010 when excluding premalignant stages and recurrent cases (Nasterlack et al. 2001). Due to
123
the increased attained age of this cohort, the number of expected cases would be higher. Hence, the standardized incidence ratio would be lower than 8.75 as reported for the previous analysis. A detailed risk assessment will be subject to another analysis with focus on exposure to aromatic amines. This lower BC incidence may reflect the longer period since carcinogenic aromatic amines have been banned from the manufacture of dyes, i.e., the decline in cases reflects the positive effects of preventive measures. How to deal with premalignant stages? Cancer screening aims to detect favorable stages before the onset of symptoms. We observed a downstaging (83 % of
Int Arch Occup Environ Health
the cases, excluding papilloma, were pTis, pTa, and pT1) in comparison with the distribution of BC in the general population (67–75 % non-muscle invasive) (Fischer et al. 1998). However, screening may lead to over-diagnosis, i.e., the detection of lesions with low premalignant potential like papillomas, the majority of which do not become malignant. Early stages are considered a step in the time course of tumor development. One case in our study had a papilloma that was not detected with markers and developed high-grade BC a few years later, which was diagnosed following elevated NMP22. The sensitivity of the marker panel increased to 86 % when papillomas were excluded from the analysis. Cystoscopy as the diagnostic gold standard may detect the tumor later than a marker. For example, genomic changes can be detected with the UroVysion assay earlier in urothelial cells exfoliated into the urine before aberrant cells can be observed with cytology or before lesions become visible in the target organ with cystoscopy. UroVysion tested more samples positive than cytology. A follow-up of these subjects is necessary to assess the positive predictive value of chromosomal aberrations. Can UroVysion in combination with NMP22 detect more bladder tumors than cytology at reasonable costs? Bladder cancer is an excellent example for the early detection of favorable stages with the non-invasive determination of tumor markers, due to the exfoliation of cancer cells or the leakage of molecular markers from the target organ into the urine. UroScreen provided evidence that the combination of a protein- and a cell-based marker can detect more cases than conventional cytology alone, a method with limited sensitivity (Sullivan et al. 2010). Out of 21 tumors observed between 2003 and 2011, 14 were detected with NMP22 or UroVysion in comparison with eight tumors with cytology. However, the laborintensive UroVysion assay overlapped with cytology, due to the preselection of atypical cells. We would suggest refraining from this selection as genomic changes may occur earlier than morphological changes of the cells would become visible. Both tests are prone to observer bias, and the costs of the FISH assay are much higher than what is currently reimbursed in Germany for cytology (170 vs. 25 €). The detection of two additional cases with UroVysion in 6,525 samples would cost 554,625 € per case. However, the strong correlation between copy number variations at three out of the four loci may likely allow a reduction in the probes and thus the development of a cheaper assay (Bonberg et al. 2013). The quantitative NMP22 assay (29 €) detected five cases in addition to cytology (40,936 € per case), but at the expense of more false-positive results, thus lowering the panel’s specificity
to 94.2 % in comparison with 98.5 % if applying cytology alone. What is the trade-off for good performance measures? In general, higher sensitivity from a panel of diagnostic markers, when compared to individual tests, can be only achieved at the expense of specificity. The trade-off between sensitivity and specificity holds for combined as well as individual tests. For example, various cutoffs have been applied to NMP22 (Abd El Gawad et al. 2005). Using 10 U/mL as recommended by the manufacturer, 3 % of our samples tested positive and six tumors were detected. The consequence of missing a subject with invasive BC favors a test with high sensitivity. A cutoff of 6 U/mL would have identified two more cases with the trade-off of four times more false-positive tests. Also, the biological features of the markers may influence their performance for BC screening. Particularly, protein-based markers may have a limited specificity as they can be involved in a variety of diseases in addition to cancer such as urinary tract infections. Nuclear matrix proteins are associated with DNA replication and RNA synthesis (Kumara-Siri et al. 1986). UroVysion seems to be more specific as it detects cancer-related genomic changes. For example, DNA loss at 9p21 has been frequently observed in cancer cells, which hosts the tumor suppressor gene p16CDKNA2 (Aravidis et al. 2012; Klorin et al. 2013; Panani et al. 2009). Diagnostic markers can only be applied in asymptomatic subjects at an acceptable rate of false-positive findings, usually not exceeding 2 % due to the invasiveness of the diagnostic workup and psychological distress (Pepe et al. 2008). Only cytology was associated with such an excellent rate but suffered from a limited sensitivity. Higher rates may be acceptable for the detection of aggressive forms like lung cancer or when monitoring patients for recurrent BC where cystoscopy is part of the medical follow-up program. Because marker validation integrated into prospective designs or surveillance programs for occupationally exposed workers is yet a bottleneck in research, performance measures from other screening studies have been rarely published (Brenner and Normolle 2007; Ransohoff 2013). So far, no randomized controlled trial which is the most rigorous study design for assessing the benefit of BC screening with markers has been performed (Larre et al. 2013). By contrast, imaging methods for the early detection of cancer were comprehensively evaluated with the National Lung Cancer Screening Trial as prominent example. Sensitivity and specificity varied by method (low-dose CT: 94.4 and 72.6 %, radiography: 59.6 and 94.1 %, respectively) (Aberle et al. 2013). Similar performance measures may likely occur in occupational settings,
123
Int Arch Occup Environ Health
for example, in lung cancer screening of asbestos-exposed workers. Voluntary screening programs with asymptomatic subjects may yield worse measures of the performance of markers than reported from clinical settings with symptomatic subjects or patients because of the smaller tumor volume at early stages and a lower compliance of healthy volunteers to attend regular screens. In addition, we observed a low compliance to repeat screens and a limited compliance for cystoscopy following a positive finding. Despite a good overall response rate in UroScreen, few men took part on an annual basis. Three out of the six interval cases had their last screen more than 1 year before diagnosis. Cytology and UroVysion were negative in all samples collected more than 12 months before diagnosis. The sensitivity of the marker panel increased to 76.5 % when restricted to prediagnostic samples from cases that underwent screens within 1 year of diagnosis. This observation is in line with a validation study on the performance of CA125 to detect ovarian cancer (Zhu et al. 2011). Does the urine status influence the test results with molecular markers? Currently, there is no consensus about the consideration of the urine status with regard to the test results. This may be due to a lack of detailed data from sufficiently large studies. Based on more than 5,000 urine samples from nondiseased participants, we confirmed that leukocytes are associated with false-positive NMP22 tests (Lotan and Shariat 2008; Sharma et al. 1999). Asymptomatic urinary tract infection can be a concomitant condition in the elderly, and a single positive NMP22 test should not initiate an invasive diagnostic procedure if supportive diagnostic data fail. The test could be repeated but we observed a low compliance when a participant was invited for a second examination (Pesch et al. 2011). On the other hand, false-negative NMP22 tests can occur in diluted urines. Frequently, urine was voided shortly before an examination although the participants were informed about the collection of spot urines. In diluted urines, NMP22 may not have reached the cutoff—as observed in two of the nondetected cases. Creatinine adjustment could be applied but may be less feasible in clinical practice than in biomonitoring studies (Todenhoefer et al. 2012). The cell content of the urine sample was a limiting factor for the cell-based tests. Samples with low creatinine content were frequently not suitable for UroVysion. However, more cells with genomic changes were detected in freshly voided urines but rarely in urines with unusually high creatinine. To address this more than tenfold variation in test positivity, we suggest collecting two urine samples, one as spot urine and a second sample after physical activity.
123
Who should receive a recommendation for a diagnostic follow-up? When applying marker panels, the rate of false-positive findings usually increases and a decision is needed to better classify these subjects into those who should repeat the tests and those who should undergo an invasive diagnostic workup. However, a consensus on cystoscopy is still lacking (Lotan et al. 2009a). It also depends on the target population whether asymptomatic subjects or BC patients are subject to screening. False-positive tests may be of less concern in BC patients where regular cystoscopy is part of the surveillance guidelines (Babjuk et al. 2011). Screening of asymptomatic volunteers requires more restrictive rules for the initiation of an invasive diagnostic workup. Subjects with abundant leukocytes should be carefully assessed before cystoscopy is recommended. For example, NMP22 concentration was highest in a dialysis patient. It is important to note that a third of the volunteers with a positive test result refused to undergo cystoscopy. We also observed some resistance to a subsequent invasive diagnostic workup when the screenee was repeatedly tested positive. This may be of relevance for subjects with urinary tract infection or inconclusive cytology. A limited specificity of diagnostic markers may require additional steps prior to an invasive diagnostic workup to minimize false-positive findings. However, complex decisions are rarely feasible in clinical routine. A recent evaluation of compliance with guidelines for BC patients revealed that only one out of 4,545 patients received all recommended measures (Chamie et al. 2011). This may imply robust tests and specialized centers with expertise in the conduct and interpretation of complex assays.
Conclusions A number of aromatic amines have been classified as human carcinogens and consequently were subjected to regulation to limit occupational exposure. In addition to efficient measures applied to improve primary prevention and in line with recommendations of the ILO, the German Social Accident Insurance offers health surveillance programs to workers with former exposure to occupational carcinogens. BC is an excellent screening target for urinary markers. Marker performance improved considerably when participants complied with the screening interval of 1 year. UroScreen revealed that protein- and cell-based markers can complement each other in the early detection of BC. However, NMP22 was associated with many false-positive tests, in particular in the presence of leukocytes. As long as UroVysion remains costly and preselects atypical cells, it will not be a useful adjunct to cytology. The relatively low
Int Arch Occup Environ Health
incidence of BC is a limiting factor when balancing benefits and cost efficiency. However, markers may be valuable for monitoring patients with BC due to the high recurrence rate of this cancer. Acknowledgments We thank Friedhelm Eberle for the careful execution and documentation of the urinalyses. This unique database allowed the exploration of the influence of the urine status on the tumor tests. Conflict of interest The study received a Grant (FB093) from the German Social Accident Insurance (DGUV), Sankt Augustin, Germany. A.S. received free assay kits from Abbott Laboratories and Matritech/Alere Ltd. G.J. received free assay reagents from Fujirebio Diagnostics, Inc. M.K. and H.W. are employees of the sponsor. Y.L. does research with Abbott Laboratories and Matritech/Alere Ltd.
Appendix UroScreen Study Group BASF SE, Department of Occupational Medicine and Health Protection, Ludwigshafen: Bernd Scheuermann, Friedhelm Eberle, Thomas Mayer, Michael Nasterlack. BG RCI, Bereich Pra¨vention, Heidelberg: Harald Wellha¨ußer, Matthias Kluckert; Organisationsdienst fu¨r nachgehende Untersuchungen (ODIN): Reinhard Detzner. Institute for Prevention and Occupational Medicine of the German Social Accident Insurance, Institute of RuhrUniversita¨t Bochum (IPA), Bochum: Dirk Taeger, Beate Pesch, Katarzyna Gawrych, Heike Bontrup, Georg Johnen, Nadine Bonberg, Judith Delbanco, Evelyn Heinze, Thomas Bru¨ning. Protein Research Unit Ruhr within Europe (PURE), Ruhr University of Bochum: Nadine Bonberg. Currenta GmbH & Co. OHG, Safety—Health Protection, Leverkusen: Gabriele Leng. Institute of Urology, Eberhard Karls University, Tu¨bingen: Gerhard Feil, Karl-Dietrich Sievert, Se´verine Banek, Christian Schwentner, Margarete Geiger, Erika Senger, Valentina Gerber, Andrea Hohneder, Gundi Beger, Ursula Kuehs, Jo¨rg Hennenlotter, Arnulf Stenzl.
References Abd El Gawad I, Moussa HS, Nasr MI, El Gemae EH, Masooud AM, Ibrahim IK, El Hifnawy NM (2005) Comparative study of NMP22, telomerase, and BTA in the detection of bladder cancer. J Egypt Natl Canc Inst 17:193–202 Aberle DR, DeMello S, Berg CD, Black WC, Brewer B, Church TR, Clingan KL, Duan F, Fagerstrom RM, Gareen IF et al (2013) Results of the two incidence screenings in the national lung screening trial. N Engl J Med 369:920–931
Aravidis C, Panani AD, Kosmaidou Z, Thomakos N, Rodolakis A, Antsaklis A (2012) Detection of numerical abnormalities of chromosome 9 and p16/CDKN2A gene alterations in ovarian cancer with fish analysis. Anticancer Res 32:5309–5313 Babjuk M, Oosterlinck W, Sylvester R, Kaasinen E, Bohle A, PalouRedorta J, Roupret M (2011) EAU guidelines on non-muscleinvasive urothelial carcinoma of the bladder, the 2011 update. Eur Urol 59:997–1008 Banek S, Schwentner C, Taeger D, Pesch B, Nasterlack M, Leng G, Gawrych K, Bonberg N, Johnen G, Kluckert M et al (2012) Prospective evaluation of fluorescence-in situ-hybridization to detect bladder cancer: results from the UroScreen-study. Urol Oncol. doi:10.1016/j.urolonc.2012.04.015 Behrens T, Bonberg N, Casjens S, Pesch B, Bru¨ning T (2013) A practical guide to epidemiological practice and standards in the identification and validation of diagnostic markers using a bladder cancer example. BBA Proteins Proteomics. doi:10.1016/ j.bbpap.2013. Article first published online: 18 July 2013 Bonberg N, Taeger D, Gawrych K, Johnen G, Banek S, Schwentner C, Sievert KD, Wellhausser H, Kluckert M, Leng G et al (2013) Chromosomal instability and bladder cancer—the UroVysionTM test in the UroScreen study. BJU Int. doi:10.1111/j.1464-410X. 2012.11666.x Brenner DE, Normolle DP (2007) Biomarkers for cancer risk, early detection, and prognosis: the validation conundrum. Cancer Epidemiol Biomarkers Prev 16:1918–1920 Chamie K, Saigal CS, Lai J, Hanley JM, Setodji CM, Konety BR, Litwin MS (2011) Compliance with guidelines for patients with bladder cancer: variation in the delivery of care. Cancer 117:5392–5401 FDA, U.S.Federal Food and Drug Administration (2000) NMP22. http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfTopic/pma/ pma.cfm?num=P940035S002. Accessed 29 May 2013 FDA, U.S.Federal Food and Drug Administration (2005) UroVysion. http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfTopic/pma/ pma.cfm?num=P030052. Accessed 29 May 2013 Fischer CG, Waechter W, Kraus S, Fuentecilla PE, Weidner W, Dudeck J (1998) Urologic tumors in the Federal Republic of Germany: data on 56,013 cases from hospital cancer registries. Cancer 82:775–783 Freudenthal RI, Stephens E, Anderson DP (1999) Determining the potential of aromatic amines to induce cancer of the urinary bladder. Int J Toxicol 18:353–359 Grossman HB, Messing E, Soloway M, Tomera K, Katz G, Berger Y, Shen Y (2005) Detection of bladder cancer using a point-of-care proteomic assay. JAMA 293:810–816 Hemstreet GP III, Yin S, Ma Z, Bonner RB, Bi W, Rao JY, Zang M, Zheng Q, Bane B, Asal N et al (2001) Biomarker risk assessment and bladder cancer detection in a cohort exposed to benzidine. J Natl Cancer Inst 93:427–436 Huber S, Schwentner C, Taeger D, Pesch B, Nasterlack M, Leng G, Mayer T, Gawrych K, Bonberg N, Pelster M et al (2012) Nuclear matrix protein-22: a prospective evaluation in a population at risk for bladder cancer. Results from the UroScreen study. BJU Int 110:699–708 Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D (2011) Global cancer statistics. CA Cancer J Clin 61:69–90 Johnen G, Gawrych K, Bontrup H, Pesch B, Taeger D, Banek S, Kluckert M, Wellhausser H, Eberle F, Nasterlack M et al (2012) Performance of survivin mRNA as a biomarker for bladder cancer in the prospective study UroScreen. PLoS One 7:e35363 Klorin G, Rozenblum E, Glebov O, Walker RL, Park Y, Meltzer PS, Kirsch IR, Kaye FJ, Roschke AV (2013) Integrated highresolution array CGH and SKY analysis of homozygous deletions and other genomic alterations present in malignant mesothelioma cell lines. Cancer Genet 206:191–205
123
Int Arch Occup Environ Health Kumara-Siri MH, Shapiro LE, Surks MI (1986) Association of the 3,5,3’-triiodo-L-thyronine nuclear receptor with the nuclear matrix of cultured growth hormone-producing rat pituitary tumor cells (GC cells). J Biol Chem 261:2844–2852 Larre S, Catto JW, Cookson MS, Messing EM, Shariat SF, Soloway MS, Svatek RS, Lotan Y, Zlotta AR, Grossman HB (2013) Screening for bladder cancer: rationale, limitations, whom to target, and perspectives. Eur Urol 63:1049–1058 Lokeshwar VB, Schroeder GL, Selzer MG, Hautmann SH, Posey JT, Duncan RC, Watson R, Rose L, Markowitz S, Soloway MS (2002) Bladder tumor markers for monitoring recurrence and screening comparison of hyaluronic acid-hyaluronidase and BTA-Stat tests. Cancer 95:61–72 Lokeshwar VB, Habuchi T, Grossman HB, Murphy WM, Hautmann SH, Hemstreet GP III, Bono AV, Getzenberg RH, Goebell P, Schmitz-Drager BJ et al (2005) Bladder tumor markers beyond cytology: international consensus panel on bladder tumor markers. Urology 66:35–63 Lotan Y, Shariat SF (2008) Impact of risk factors on the performance of the nuclear matrix protein 22 point-of-care test for bladder cancer detection. BJU Int 101:1362–1367 Lotan Y, Svatek RS, Sagalowsky AI (2006) Should we screen for bladder cancer in a high-risk population? A cost per life-year saved analysis. Cancer 107:982–990 Lotan Y, Capitanio U, Shariat SF, Hutterer GC, Karakiewicz PI (2009a) Impact of clinical factors, including a point-of-care nuclear matrix protein-22 assay and cytology, on bladder cancer detection. BJU Int 103:1368–1374 Lotan Y, Elias K, Svatek RS, Bagrodia A, Nuss G, Moran B, Sagalowsky AI (2009b) Bladder cancer screening in a high risk asymptomatic population using a point of care urine based protein tumor marker. J Urol 182:52–57 Nasterlack M, Scheuermann B, Messerer P, Pallapies D, Zober A (2001) Harnblasenkrebs in einem Risikokollektiv: Klinische und epidemiologische Aspekte. Symp Med 12:17–19 Panani AD, Maliaga K, Babanaraki A, Bellenis I (2009) Numerical abnormalities of chromosome 9 and p16CDKN2A gene deletion detected by FISH in non-small cell lung cancer. Anticancer Res 29:4483–4487 Pepe MS, Feng Z, Janes H, Bossuyt PM, Potter JD (2008) Pivotal evaluation of the accuracy of a biomarker used for classification
123
or prediction: standards for study design. J Natl Cancer Inst 100:1432–1438 Pesch B, Nasterlack M, Eberle F, Bonberg N, Taeger D, Leng G, Feil G, Johnen G, Ickstadt K, Kluckert M et al (2011) The role of haematuria in bladder cancer screening among men with former occupational exposure to aromatic amines. BJU Int 108:546–552 Ransohoff DF (2013) Cultivating cohort studies for observational translational research. Cancer Epidemiol Biomarkers Prev 22:481–484 Rehn L (1895) Blasengeschwu¨lste bei Fuchsin-Arbeitern. Archiv fu¨r Klinische Chirurgie 50:588–600 Sauter G, Algaba F, Amin M, Busch C, Cheville J, Gasser T, Grignon D, Hofstaedter F, Lopez-Beltran A, Epstein J (2004) Noninvasive urothelial neoplasias: WHO classification of noninvasive papillary urothelial tumors. In: Eble JN, Sauter G, Epstein JI, Sesterhenn IA (eds) World Health Organization classification of tumors. Pathology and genetics of tumors of the urinary system and male genital organs, 3rd edn. IARC Press, Lyon Shariat SF, Karam JA, Lotan Y, Karakiewizc PI (2008) Critical evaluation of urinary markers for bladder cancer detection and monitoring. Rev Urol 10:120–135 Sharma S, Zippe CD, Pandrangi L, Nelson D, Agarwal A (1999) Exclusion criteria enhance the specificity and positive predictive value of NMP22 and BTA stat. J Urol 162:53–57 Sullivan P, Chan J, Levin M, Rao J (2010) Urine cytology and adjunct markers for detection and surveillance of bladder cancer. Am J Transl Res 2:412–440 Todenhoefer T, Hennenlotter J, Witstruk M, Gakis G, Aufderklamm S, Kuehs U, Stenzl A, Schwentner C (2012) Influence of renal excretory function on the performance of urine based markers to detect bladder cancer. J Urol 187:68–73 van Rhijn BW, van der Poel HG, Van der Kwast TH (2005) Urine markers for bladder cancer surveillance: a systematic review. Eur Urol 47:736–748 Vineis P, Pirastu R (1997) Aromatic amines and cancer. Cancer Causes Control 8:346–355 Zhu CS, Pinsky PF, Cramer DW, Ransohoff DF, Hartge P, Pfeiffer RM, Urban N, Mor G, Bast RC Jr, Moore LE et al (2011) A framework for evaluating biomarkers for early detection: validation of biomarker panels for ovarian cancer. Cancer Prev Res (Phila) 4:375–383
