Background
Patients with non-muscle invasive bladder cancer (NMIBC) harbour a significant risk of tumor recurrence and progression [
1]. Several risk stratifications have been developed to predict the risk of recurrence or progression based on clinical and pathologic parameters [
2]. However, these tools have only limited value in predicting the patients’ individual risk. Therefore, improved risk stratification is urgently needed. Due to the lack of reliable tools for prognosis of patients with NMIBC, the optimal follow-up of these patients is discussed controversially. White light cystoscopy remains the gold standard for surveillance of patients after NMIBC. The main limitations of cystoscopy are its limited sensitivity for flat lesions and its character as an invasive procedure potentially leading to significant discomfort for the patients [
3,
4]. Urine cytology is also recommended as standard in the follow-up of patients with NMIBC, as it is a non-invasive procedure with the potential to detect flat lesions not visible in cystoscopy [
5]. However, its sensitivity is satisfactory only for high grade tumors or
carcinoma in situ. Newer markers such as fluorescence-in-situ hybridization (FISH), immunocytology (uCyt+) or Nuclear matrix protein 22 (NMP22) have shown increased sensitivitity compared to cytology [
6,
7]. However, their specificity has been reported to be lower compared to cytology in most studies. For some of these markers, particularly the UroVysion FISH test previous studies have shown that a positive marker might precede visual or histologic detection of a tumor recurrence. Hence, some tests may be capable of detecting molecular changes associated with tumor recurrence earlier than cystoscopy [
8,
9]. The clinical implications and the optimal management of these patients with negative cystoscopy or biopsy and concomitantly positive markers remain to be defined.
To date only few data exist on the predictive values of multiple urine markers in this negative-cystoscopy setting. The present study is the first to address this issue concerning the four most widely available urine markers (Cytology, FISH, uCyt + and NMP22).
Discussion
The use of urine markers in the surveillance of patients with NMIBC has increased significantly in the last decade, although current guidelines give clear recommendations only for cytology. Urine markers are considered to be a valuable adjunct to cystoscopy [
17,
18]. However, the optimal work-up of patients with negative cystoscopy and positive urine markers - particularly for newer markers with decreased specificity compared to cytology - has not been clearly defined yet. It has been frequently discussed whether the presence of a positive marker in the absence of obvious changes in the bladder mucosa may indicate the presence of a non-visible tumor. Others even consider predicting the development of a tumor in the near future. Several reports indicate that anticipatory positive urine markers may indeed precede clinical tumor recurrence. The phenomenon of anticipatory positive markers has been firstly described by Steiner et al, who assessed the feasibility of microsatellite analysis in the follow up of 21 patients with NMIBC. In their cohort, two patients were positive for urine microsatellite analysis four and six months prior to cystoscopic detection of tumors [
19]. A comparison of multiple urine markers and their combination has not been performed yet with regards to their anticipatory role. The aim of the present study was to compare the oncologic outcomes of patients with positive and negative cytology, FISH, uCyt+ and NMP22 at time of negative cystoscopy. Furthermore, we aimed to assess the impact of these markers on recurrence and progression when used in combinations.
In univariable analysis we observed for all markers to be highly predictive for recurrence and progression. Of note, patients with positive markers were at higher risk for recurrence even 1-2 years after urine marker testing and negative cystoscopy. This implies, that the tests might be able to display premalignant changes that are not visible macroscopically. Interestingly, in the group of patients having negative cytology, the only marker that was associated with increased risk of recurrence was NMP22. This indicates that both FISH and immunocytology markedly overlapped with cytology with regards to prediction. When combining urine markers, we observed that the more markers are positive in different combinations, the higher is the risk for recurrence and progression. For various combinations including NMP22, the risk of having disease recurrence and progression within 24 months was as low as 9.5-13.5% and 2.8-5.4% for patients with both tests negative (depending on the combination). Of note, a negative 4-test combination was not associated with a further decrease in the probability of recurrence and progression compared to the 2-test-combination with the highest negative predictive value (when considering recurrence as an event). Therefore, the use of more than two markers does not seem to provide additional benefit with regards to prediction of outcome.
The observation, that positive urine markers may precede visible tumor recurrence is in accordance with previous studies. In a cohort of 1114 patients tested by urine cytology, the anticipatory positive rate of cytology was 44% after a median time of 15 months [
9]. Compared to our study, the rate of patients having a positive cytology but no positive histology within one year of follow-up was clearly lower. This finding may be related to the fact that cytology is strongly dependant on the cytopathologist [
20].
UroVysion FISH has been observed to be positive before visible tumor recurrence in various studies. In a study by Yoder et al., 56 of 250 patients with atypical or negative cytology were positive for FISH and had no visible tumor recurrence at initial cystoscopy. In 35 of these patients (62.5%), tumor recurrence developed during the median follow-up of 23 months. Similar results were observed in a study including 68 patients under surveillance for NMIBC having negative cytology and cystoscopy at time of inclusion. During the median follow-up of 13.5 months, 45% of patients with positive UroVysion developed recurrence vs. 12.5% with normal UroVysion test. The proportion of patients with positive UroVysion at time of inclusion was considerably high (62.5%), given the fact that all were negative for cystoscopy at time of testing [
21]. These findings are in contrast to our study, where in the group of patients with negative cytology, the rates of recurrence did not differ significantly between FISH positive and negative patients. This might be also caused by other criteria used for positive cytology. In a study from Italy 75 patients during follow-up of NMIBC were divided into patients with low molecular grade (changes in 9p21 or chromosome 3) and high-molecular grade (gains of chromosome 7 or 17) FISH results. Those patients with high molecular grade had a significantly shorter time to recurrence. In this study, no substratification was performed for cytology positive and negative patients [
22]. However, the criteria used in this study were different compared to ours. It is broadly accepted that the interpretation criteria might strongly influence the prognostic value of UroVysion results. In a study including 138 patients with negative cystoscopy during follow-up for NMIBC, the UroVysion criteria proposed by the manufacturer were not predictive for recurrence. In contrast, criteria based on previous evidence suggesting that rare tetraploidic cells have a less strong diagnostic value than other aberrations, showed a significant association with disease recurrence [
23,
24]. Another test based on detection of genomic alterations which has been demonstrated to predict a later tumor recurrence is microsatellite analysis (MA). In a study by van Rhjin et al., MA was able to detect 75% of tumor recurrences. In the group of patients with positive MA and negative cystoscopy, 55% experienced recurrence within six months vs. 11.6% in patients with negative MA [
25]. Similar results were obtained by van der Aa et al., who prospectitvely assessed the predictive value of MA in a longitudinal analysis for 458 series with persistent MA results. They observed, that patients with persistent positive MA analysis had a 83% risk of recurrence after two years vs. 22% in patients with persistent negative MA analysis. Positive MA series preceded recurrences by one to 24 months [
26].
Analysis of FGFR3 mutations in voided urine samples of patients with history of FGFR3 mutant tumors was performed in a study by Zuiverloon et al. They observed, that that a single positive FGFR3 test was associated with a three times higher risk of a recurrence. In patients with consecutive FGFR3-positive urine samples the risk of developing a recurrence within 39 months was 90% [
27].
For immunocytology, no report exists so far on anticipatory positive tests. In a study from Montreal including 109 patients during surveillance of BC, immunocytology results were correlated with tumor presence within 12 months [
28]. However, the study, which showed a proportion of patients with tumor during 12 months of 76%, does not provide detailed information on when the recurrences occurred and how immunocytology predicted recurrences in patients with a negative cystoscopy at study inclusion. Similar to FISH, we observed no difference in recurrence free survival between patients with negative cytology and positive or negative immunocytology in our study. This might be due to a high percentage of patients (68.4%) showing an overlap of results of cytology and immunocytology.
Interestingly, NMP22 was the only marker being strongly predictive for recurrence in cytology negative patients. This might be explained by the different features covered by both tests. In our study, it was the test with the lowest concordance with cytology (43.0% of patients showed different results for cytology and NMP22). Such a strong correlation with risk of recurrence has not been observed in literature yet. However, the fact that NMP22 might be a test offering more additional information than other markers when used in combination with cytology has also been observed before [
28,
29]. Besides NMP22, other protein based urine markers have been also investigated for their potential in predicting recurrences in patients undergoing surveillance of NMIBC. In a study by Sanchez-Carbayo et al, patients under surveillance received serial testing for Urinary Bladder Cancer test (UBC), CYFRA-21-1 and NMP22. In the group of 65 patients with persistent negative markers during the study, only four patients developed recurrence within the one year follow-up of the study. The authors concluded, that urinary markers should be considered as adjuncts enabling individualized cystoscopy intervals during NMIBC surveillance [
30].
The present study is the first study addressing the question, how combinations of multiple markers may help to improve prediction of recurrence and progression when assessed at time of negative cystoscopy. Although we observed that patients having all tests positive in a 4-marker-combination are at higher risk for recurrence than patients with less markers positive, the identification of patients with low risk of recurrence and progression was not improved significantly when using four instead of two markers. When discussing urine markers as potential tool to individualize cystoscopy intervals during BC surveillance, patients and urologists expect a high negative predictive value (NPV) for this test with regards to prediction of recurrence and/or progression. Therefore, the use of more than two markers has to be questioned in this context, as it does not seem to improve the NPV compared to the 2-test-combinations with high NPV. Furthermore, the use of multiple markers is associated with a significant increase in costs.
The limitations of our study include the size and the heterogeneity of the cohort. Although we controlled for this bias in multivariate analysis, the inclusion of patients with different intervals between last evidence of tumor and urine marker testing might have influenced the results. Furthermore, we cannot exclude that the results of the urine markers influenced cystoscopy intervals as treating urologist were not blinded to the test results. However, the inclusion of patients with a follow-up of at least 24 months and the performance of tests which took the presence of tumor within 24 months as an endpoint (regardless of the time to recurrence) may partially control for this bias. As only a part of the patients underwent mapping biopsies during cystoscopy, it cannot be completely ruled out that carcinoma in situ was present at time of initial cystoscopic assessment. Moreover, the sensitivity of upper tract imaging for detection of upper urinary tract urothelial carcinoma is limited. To exclude presence of upper tract tumors, a ureteroscopic assessment of all patients would have been necessary. However, none of the patients presented clinical symptoms of upper tract urothelial carcinoma within the follow-up of two years.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
TT contributed to study design, data acquisition, statistical analysis and writing of the manuscript. JH contributed to data acquisition, statistical analysis and writing of the manuscript. PG contributed to data acquisition. SM contributed to data acquisition. UK contributed to data acquisition and writing of the manuscript. ME contributed to data acquisition. SA contributed to writing and review of the manuscript. SB contributed to writing and review of the manuscript. NH contributed to revision and review of the manuscript. SR contributed to data acquisition and review of the manuscript. GG contributed to study design and review of the manuscript. AS contributed to study design and writing of the manuscript. CS was responsible for study design, statistical analysis and writing of the manuscript. All authors read and approved the final manuscript.