REVIEWS Developing proteomic biomarkers for bladder cancer: towards clinical application Maria Frantzi, Agnieszka Latosinska, Leif Flühe, Marie C. Hupe, Elena Critselis, Mario W. Kramer, Axel S. Merseburger, Harald Mischak and Antonia Vlahou Abstract | Clinical use of proteomic biomarkers has the potential to substantially improve the outcomes of patients with bladder cancer. An unmet clinical need evidently exists for noninvasive biomarkers, which might enable improvements in both the diagnosis and prognosis of patients with bladder cancer, as well as improved monitoring of patients for the presence of recurrence. Urine is considered the optimal noninvasive source of proteomic biomarkers in patients with bladder cancer. Currently, a number of single-protein biomarkers have been detected in urine and tissue using a variety of proteomic techniques, each having specific conceptual considerations and technical implications. Promising preclinical data are available for several of these proteins; however, the combination of single urinary proteins into multimarker panels might better encompass the molecular heterogeneity of bladder cancer within this patient population, and prove more effective in clinical use. Frantzi, M. et al. Nat. Rev. Urol. advance online publication 26 May 2015; doi:10.1038/nrurol.2015.100
Introduction
Biotechnology Division, Biomedical Research Foundation Academy of Athens, 4 Soranou Ephessiou Street, 115 27 Athens, Greece (M.F., A.L., E.C., A.V.). Mosaiques Diagnostics GmbH, Rotenburger Strasse 20, 30659 Hannover, Germany (L.F.). Department of Urology and Urological Oncology, Hannover Medical School, Carl‑Neuberg-Strasse 1, 30625 Hannover, Germany (M.C.H., M.W.K., A.S.M.). BHF Glasgow Cardiovascular Research Centre, University of Glasgow, 126 University Avenue, Glasgow G12 8TA, UK (H.M.). Correspondence to: M.F. [email protected]
Bladder cancer is the second most frequent cause of mortality among cancers of the genitourinary system with approximately 429,000 newly diagnosed cases and 165,000 attributable deaths worldwide in 2012. 1 Bladder cancer is the sixth most common of all malignancies in men. 1 Among the global population, the age-standardized rate of bladder cancer incidence is highest in North America, Europe and North Africa.2 Smoking remains the main risk factor for bladder cancer, accounting for almost 50% of all cases.3 Approximately 90% of malignant bladder tumours are urothelial cell carcinomas, which originate in the epithelium. 80% of all neoplasms are classified as non- muscle-invasive bladder cancer (NMIBC) of stages pTa/ pT1 or carcinoma in situ (CIS) at the time of initial presentation, based on the tumour size and the extent of infiltration into the bladder urothelium; the remaining 20% of patients present with muscle-invasive bladder cancer (MIBC) of stages ≥pT2. 4 Patients with MIBC generally have a poor prognosis with more than 50% of patients succumbing to their disease within 5 years of diagnosis.5 Over 70% of patients with NMIBC will have at least one disease recurrence after successful initial treatment.6,7 The highest rates of recurrence are observed in patients with high-grade T1 tumours (pT1G3), which also have the highest probability of progression (29%) 5 years after initial diagnosis.6,8
Competing Interests M.F. is employed by Mosaiques Diagnostics as part of an EID (Industrial-Academia) Marie Curie Action. L.F. is employed by Mosaiques Diagnostics. H.M. is the founder and co-owner of Mosaiques Diagnostics. The other authors declare no competing interests.
Patients with NMIBC can be stratified into three main groups with respect to the risk of disease recurrence or progression (low, intermediate or high risk).6 The currently available clinical means for initial disease diagnosis involve cystoscopy, which is both unpleasant and invasive. As an alternative to the invasive cystoscopy, urinary cytology has been applied, with suboptimal sensitivity, especially for low-grade tumours.9 Cystoscopy, is the current ‘gold standard’ diagnostic procedure; however, use of cystoscopy has been reported to miss papillary tumours (Ta and T1),10 as well as high-grade CIS lesions.11 In a meta-analysis,12 71% (49–93%) sensitivity and 72% (47–96%) specificity were reported in studies using white-light cystoscopy to diagnose bladder cancer, while sensitivity of 44% (38–51%) and specificity of 96% (94– 98%) were reported in studies using urinary cytology. Moreover, owing to the high rates of recurrence and progression associated with bladder cancer 7 current guidelines on the monitoring of patients for recurrence include use of cystoscopy and voided urine cytology, at regular intervals.6 This need for frequent follow-up monitor ing encumbers the resources of health-care systems and might adversely affect patients’ compliance to monitoring.13 Accurate, noninvasive tests for both the early diagnosis and detection of recurrent bladder cancer and/or its progression are, therefore, required.14 The introduction of noninvasive tests would reduce the need for surveillance cystoscopies, and/or increase the accuracy of the findings of diagnostic procedures. Evaluations of the risks versus benefits and other considerations concerning the clinical use of noninvasive biomarkers, especially for the detection of NMIBC, have been reported elsewhere in patients with high-risk NMIBC,15 as well as in those with low-risk or intermediate-risk NMIBC.16
NATURE REVIEWS | UROLOGY
ADVANCE ONLINE PUBLICATION | 1 © 2015 Macmillan Publishers Limited. All rights reserved
REVIEWS Currently available urine tests
Key points ■■ An unmet clinical need exists for accurate, noninvasive assays, which enable the accurate diagnosis of bladder cancer and monitoring of patients for recurrence ■■ Urine is the most appropriate bodily fluid for biomarker research owing to its noninvasive and easy collection, stability of samples and proximity to bladder tumours ■■ Proteomics platforms have the advantages of providing large datasets with numerous putative biomarker candidates at high resolution; these technologies might also potentially enable validation of biomarker candidates ■■ Multiple biomarker panels are likely to be more effective than single biomarkers, owing to the high level of disease heterogeneity observed among patients with bladder cancer ■■ Multiple proteomic biomarkers for bladder cancer have already been discovered; however, a lack of validation in the appropriate patient populations and in specific contexts of use precludes clinical implementation
Box 1 | Clinical needs for, and potential applications of protein biomarkers Diagnosis and monitoring ■■ Early, noninvasive detection ■■ Increased sensitivity for CIS ■■ Detection of tumour recurrence ■■ Risk-adapted follow-up monitoring Prognosis ■■ Disease-specific survival ■■ Overall survival ■■ Progression of NMIBC (including CIS) to MIBC Recurrence ■■ Prediction of treatment response ■■ Response to intravesical chemotherapy for low to intermediate-risk NMIBC ■■ Response to intravesical BCG treatment for intermediate to high-risk NMIBC or CIS ■■ Selection of highest-risk patients for earlier or more aggressive interventions ■■ Responses of patients with MIBC to (neo)adjuvant chemotherapy ■■ Responses of patients with metastatic bladder cancer to chemotherapy Therapeutic interventions ■■ Identification of new targets for drug development Abbreviations: CIS, carcinoma in situ; MIBC, muscle-invasive bladder cancer; NMIBC, non‑muscle-invasive bladder cancer.
In addition to disease diagnosis, patients with bladder cancer have a variety of clinical needs and priorities (Boxes 1 and 2). The clinical utility of newly discovered biomarkers depends upon the added value these provide, beyond currently available standards of care. The clinical potential and challenges in implementing biomarkers in the management of patients who have bladder cancer have been summarized elsewhere, addressing different aspects according to the intended purpose: diagnosis, 17 prognosis 18,19 or prediction of treatment response.20 Guided by the clinical needs of patients with bladder cancer, the main focus of this article is on noninvasive proteomic biomarkers of primary and recurrent bladder cancer. In an effort to reflect advances in the field of prognostic and predictive protein biomarkers, a brief overview on this topic is also provided. For diagnostic purposes, use of urine-based assays and methodologies is the most effective approach, whereas for prognosis and/ or prediction of outcomes, tissue biomarkers appear most promising.
In an effort to overcome the need for cystoscopy, several urine-based tests have received FDA approval: immuno assays to detect urinary proteins, such as bladder-cancerassociated antigens (BTA TRAK®, BTA stat®, Polymedco, NY, USA)21 and nuclear matrix protein NMP22®(Alere, MA, USA);22 an immunocytofluorescence-based test (ImmunoCyt™/uCyt+™, Scimedx, NJ, USA);23 and a fluorescence in situ hybridization-based assay (UroVysion®, Abbott laboratories, IL, USA).24 The initial overall performance of these FDA approved tests was encouraging. Unfortunately this performance has not always been reproduced when the assays are evaluated in independent patient populations (Table 1).25,26 This finding is most evident from the results of the UroScreen study, wherein NMP22®was used to screen a population with a high risk of bladder cancer owing to previous chemical exposures; the specificity of NMP22® was found to be only 29%.25 A similar observation was evident for the UroVysion® assay, in which the reported sensitivity was only 45%.26 ImmunoCyt™ also proved to be less effective when tested in 506 patients with atypical urinary cytology, resulting in a specificity of 49%.27 These results suggest that the overall performance of currently available, noninvasive urinary tests in real-life patient populations has not yet proven sufficiently effective to either reduce the use of cystoscopy or improve its diagnostic performance.9,25–27 Thus, the clinical utility of the FDAapproved biomarkers for diagnosis of bladder cancer remains questionable and a substantial need for implementation of better biomarkers into routine clinical practice remains.
Proteomic platforms
Proteomic technologies are the ideal techniques for the detection and investigation of biomarker candidates, owing to the high sensitivity and analytical performance that can be achieved and the ability to generate large datasets through identification of large and everincreasing numbers of proteins, including peptides or protein fragments. The proteomics datasets are also indicative of the absolute or relative abundance of proteins,28 and these technolgies also have the potential to investigate both intermolecular interactions29,30 and post‑translational modifications.31 In an effort to improve current clinical practices, proteomics approaches have been widely applied to the analysis of a variety of clinical specimens such as urine, blood or tissue, with an overall aim of identifying disease-specific biomarkers or molecular signatures (Figure 1, Table 2). More detailed descriptions of the different proteomics platforms used in the context of clinical proteomics are available elsewhere.32,33 Prior to the selection of the analytical strategy, the platform and protocols to be used must be well-characterized, and the analytical variability and measurement precision of such approaches must be established.34 In general, proteomics platforms (such as mass spectrometry) have been used for global proteomic profiling as well as for targeted analyses (such as multiple-reaction monitoring
2 | ADVANCE ONLINE PUBLICATION
www.nature.com/nrurol © 2015 Macmillan Publishers Limited. All rights reserved
REVIEWS Box 2 | Priorities of biomarker research NMIBC and/or MIBC ■■ Context of use: diagnosis of local recurrence ■■ Added value: guide cystoscopy; reduce the use of surveillance cystoscopies High-risk NMIBC ■■ Context of use: prediction of progression and treatment responses ■■ Added value: guidance of treatment selection, enabling adjustments of the aggressiveness of therapeutic modalities MIBC ■■ Context of use: prediction of treatment response ■■ Added value: extension of disease survival Abbreviations: MIBC, muscle-invasive bladder cancer; NMIBC, non-muscle-invasive bladder cancer.
[MRM], or immune-based techniques). Moreover, the high complexity of the human proteome imposes the need for fractionation before mass spectrometry analysis using two-dimensional gel electrophoresis (2DE), liquid chromatography or capillary electrophoresis.
Proteomics studies in bodily fluids
Several promising potential biomarkers have been discovered in bodily fluids including plasma and urine, using high-throughput proteomics approaches. These have the advantage of enabling noninvasive clinical measurements in patients.
Blood biomarker detection Four published studies35–38 have investigated proteomic biomarkers of bladder cancer in plasma or serum, yielding several candidate biomarkers, such as protein S100-A8 (also known as calgranulin-A), protein S100-A9 (also known as calgranulin-B),35 α-1-acid glycoprotein 1 (AGP1), 36 carbonic anhydrase 1, haptoglobin and leucine-rich α‑2-glycoprotein (LRG).37 Classification models have also been generated based on the selected discriminatory peaks between bladder cancer patients and healthy individuals.38 Preoperative and postoperative sera from two patients with bladder cancer were analysed using 2DE and revealed decreased expression of 68 protein spots in postoperative samples compared to the preoperative sera.35 Further evaluation of protein S100-A8 and protein S100-A9 levels was conducted in 77 tumour-tissue specimens using immunohisto chemistry. A scoring system was adopted based on the staining intensity and the percentage of positively stained cells. Significantly higher levels of protein S100-A8 were correlated with muscle invasion (P
Introduction
Biotechnology Division, Biomedical Research Foundation Academy of Athens, 4 Soranou Ephessiou Street, 115 27 Athens, Greece (M.F., A.L., E.C., A.V.). Mosaiques Diagnostics GmbH, Rotenburger Strasse 20, 30659 Hannover, Germany (L.F.). Department of Urology and Urological Oncology, Hannover Medical School, Carl‑Neuberg-Strasse 1, 30625 Hannover, Germany (M.C.H., M.W.K., A.S.M.). BHF Glasgow Cardiovascular Research Centre, University of Glasgow, 126 University Avenue, Glasgow G12 8TA, UK (H.M.). Correspondence to: M.F. [email protected]
Bladder cancer is the second most frequent cause of mortality among cancers of the genitourinary system with approximately 429,000 newly diagnosed cases and 165,000 attributable deaths worldwide in 2012. 1 Bladder cancer is the sixth most common of all malignancies in men. 1 Among the global population, the age-standardized rate of bladder cancer incidence is highest in North America, Europe and North Africa.2 Smoking remains the main risk factor for bladder cancer, accounting for almost 50% of all cases.3 Approximately 90% of malignant bladder tumours are urothelial cell carcinomas, which originate in the epithelium. 80% of all neoplasms are classified as non- muscle-invasive bladder cancer (NMIBC) of stages pTa/ pT1 or carcinoma in situ (CIS) at the time of initial presentation, based on the tumour size and the extent of infiltration into the bladder urothelium; the remaining 20% of patients present with muscle-invasive bladder cancer (MIBC) of stages ≥pT2. 4 Patients with MIBC generally have a poor prognosis with more than 50% of patients succumbing to their disease within 5 years of diagnosis.5 Over 70% of patients with NMIBC will have at least one disease recurrence after successful initial treatment.6,7 The highest rates of recurrence are observed in patients with high-grade T1 tumours (pT1G3), which also have the highest probability of progression (29%) 5 years after initial diagnosis.6,8
Competing Interests M.F. is employed by Mosaiques Diagnostics as part of an EID (Industrial-Academia) Marie Curie Action. L.F. is employed by Mosaiques Diagnostics. H.M. is the founder and co-owner of Mosaiques Diagnostics. The other authors declare no competing interests.
Patients with NMIBC can be stratified into three main groups with respect to the risk of disease recurrence or progression (low, intermediate or high risk).6 The currently available clinical means for initial disease diagnosis involve cystoscopy, which is both unpleasant and invasive. As an alternative to the invasive cystoscopy, urinary cytology has been applied, with suboptimal sensitivity, especially for low-grade tumours.9 Cystoscopy, is the current ‘gold standard’ diagnostic procedure; however, use of cystoscopy has been reported to miss papillary tumours (Ta and T1),10 as well as high-grade CIS lesions.11 In a meta-analysis,12 71% (49–93%) sensitivity and 72% (47–96%) specificity were reported in studies using white-light cystoscopy to diagnose bladder cancer, while sensitivity of 44% (38–51%) and specificity of 96% (94– 98%) were reported in studies using urinary cytology. Moreover, owing to the high rates of recurrence and progression associated with bladder cancer 7 current guidelines on the monitoring of patients for recurrence include use of cystoscopy and voided urine cytology, at regular intervals.6 This need for frequent follow-up monitor ing encumbers the resources of health-care systems and might adversely affect patients’ compliance to monitoring.13 Accurate, noninvasive tests for both the early diagnosis and detection of recurrent bladder cancer and/or its progression are, therefore, required.14 The introduction of noninvasive tests would reduce the need for surveillance cystoscopies, and/or increase the accuracy of the findings of diagnostic procedures. Evaluations of the risks versus benefits and other considerations concerning the clinical use of noninvasive biomarkers, especially for the detection of NMIBC, have been reported elsewhere in patients with high-risk NMIBC,15 as well as in those with low-risk or intermediate-risk NMIBC.16
NATURE REVIEWS | UROLOGY
ADVANCE ONLINE PUBLICATION | 1 © 2015 Macmillan Publishers Limited. All rights reserved
REVIEWS Currently available urine tests
Key points ■■ An unmet clinical need exists for accurate, noninvasive assays, which enable the accurate diagnosis of bladder cancer and monitoring of patients for recurrence ■■ Urine is the most appropriate bodily fluid for biomarker research owing to its noninvasive and easy collection, stability of samples and proximity to bladder tumours ■■ Proteomics platforms have the advantages of providing large datasets with numerous putative biomarker candidates at high resolution; these technologies might also potentially enable validation of biomarker candidates ■■ Multiple biomarker panels are likely to be more effective than single biomarkers, owing to the high level of disease heterogeneity observed among patients with bladder cancer ■■ Multiple proteomic biomarkers for bladder cancer have already been discovered; however, a lack of validation in the appropriate patient populations and in specific contexts of use precludes clinical implementation
Box 1 | Clinical needs for, and potential applications of protein biomarkers Diagnosis and monitoring ■■ Early, noninvasive detection ■■ Increased sensitivity for CIS ■■ Detection of tumour recurrence ■■ Risk-adapted follow-up monitoring Prognosis ■■ Disease-specific survival ■■ Overall survival ■■ Progression of NMIBC (including CIS) to MIBC Recurrence ■■ Prediction of treatment response ■■ Response to intravesical chemotherapy for low to intermediate-risk NMIBC ■■ Response to intravesical BCG treatment for intermediate to high-risk NMIBC or CIS ■■ Selection of highest-risk patients for earlier or more aggressive interventions ■■ Responses of patients with MIBC to (neo)adjuvant chemotherapy ■■ Responses of patients with metastatic bladder cancer to chemotherapy Therapeutic interventions ■■ Identification of new targets for drug development Abbreviations: CIS, carcinoma in situ; MIBC, muscle-invasive bladder cancer; NMIBC, non‑muscle-invasive bladder cancer.
In addition to disease diagnosis, patients with bladder cancer have a variety of clinical needs and priorities (Boxes 1 and 2). The clinical utility of newly discovered biomarkers depends upon the added value these provide, beyond currently available standards of care. The clinical potential and challenges in implementing biomarkers in the management of patients who have bladder cancer have been summarized elsewhere, addressing different aspects according to the intended purpose: diagnosis, 17 prognosis 18,19 or prediction of treatment response.20 Guided by the clinical needs of patients with bladder cancer, the main focus of this article is on noninvasive proteomic biomarkers of primary and recurrent bladder cancer. In an effort to reflect advances in the field of prognostic and predictive protein biomarkers, a brief overview on this topic is also provided. For diagnostic purposes, use of urine-based assays and methodologies is the most effective approach, whereas for prognosis and/ or prediction of outcomes, tissue biomarkers appear most promising.
In an effort to overcome the need for cystoscopy, several urine-based tests have received FDA approval: immuno assays to detect urinary proteins, such as bladder-cancerassociated antigens (BTA TRAK®, BTA stat®, Polymedco, NY, USA)21 and nuclear matrix protein NMP22®(Alere, MA, USA);22 an immunocytofluorescence-based test (ImmunoCyt™/uCyt+™, Scimedx, NJ, USA);23 and a fluorescence in situ hybridization-based assay (UroVysion®, Abbott laboratories, IL, USA).24 The initial overall performance of these FDA approved tests was encouraging. Unfortunately this performance has not always been reproduced when the assays are evaluated in independent patient populations (Table 1).25,26 This finding is most evident from the results of the UroScreen study, wherein NMP22®was used to screen a population with a high risk of bladder cancer owing to previous chemical exposures; the specificity of NMP22® was found to be only 29%.25 A similar observation was evident for the UroVysion® assay, in which the reported sensitivity was only 45%.26 ImmunoCyt™ also proved to be less effective when tested in 506 patients with atypical urinary cytology, resulting in a specificity of 49%.27 These results suggest that the overall performance of currently available, noninvasive urinary tests in real-life patient populations has not yet proven sufficiently effective to either reduce the use of cystoscopy or improve its diagnostic performance.9,25–27 Thus, the clinical utility of the FDAapproved biomarkers for diagnosis of bladder cancer remains questionable and a substantial need for implementation of better biomarkers into routine clinical practice remains.
Proteomic platforms
Proteomic technologies are the ideal techniques for the detection and investigation of biomarker candidates, owing to the high sensitivity and analytical performance that can be achieved and the ability to generate large datasets through identification of large and everincreasing numbers of proteins, including peptides or protein fragments. The proteomics datasets are also indicative of the absolute or relative abundance of proteins,28 and these technolgies also have the potential to investigate both intermolecular interactions29,30 and post‑translational modifications.31 In an effort to improve current clinical practices, proteomics approaches have been widely applied to the analysis of a variety of clinical specimens such as urine, blood or tissue, with an overall aim of identifying disease-specific biomarkers or molecular signatures (Figure 1, Table 2). More detailed descriptions of the different proteomics platforms used in the context of clinical proteomics are available elsewhere.32,33 Prior to the selection of the analytical strategy, the platform and protocols to be used must be well-characterized, and the analytical variability and measurement precision of such approaches must be established.34 In general, proteomics platforms (such as mass spectrometry) have been used for global proteomic profiling as well as for targeted analyses (such as multiple-reaction monitoring
2 | ADVANCE ONLINE PUBLICATION
www.nature.com/nrurol © 2015 Macmillan Publishers Limited. All rights reserved
REVIEWS Box 2 | Priorities of biomarker research NMIBC and/or MIBC ■■ Context of use: diagnosis of local recurrence ■■ Added value: guide cystoscopy; reduce the use of surveillance cystoscopies High-risk NMIBC ■■ Context of use: prediction of progression and treatment responses ■■ Added value: guidance of treatment selection, enabling adjustments of the aggressiveness of therapeutic modalities MIBC ■■ Context of use: prediction of treatment response ■■ Added value: extension of disease survival Abbreviations: MIBC, muscle-invasive bladder cancer; NMIBC, non-muscle-invasive bladder cancer.
[MRM], or immune-based techniques). Moreover, the high complexity of the human proteome imposes the need for fractionation before mass spectrometry analysis using two-dimensional gel electrophoresis (2DE), liquid chromatography or capillary electrophoresis.
Proteomics studies in bodily fluids
Several promising potential biomarkers have been discovered in bodily fluids including plasma and urine, using high-throughput proteomics approaches. These have the advantage of enabling noninvasive clinical measurements in patients.
Blood biomarker detection Four published studies35–38 have investigated proteomic biomarkers of bladder cancer in plasma or serum, yielding several candidate biomarkers, such as protein S100-A8 (also known as calgranulin-A), protein S100-A9 (also known as calgranulin-B),35 α-1-acid glycoprotein 1 (AGP1), 36 carbonic anhydrase 1, haptoglobin and leucine-rich α‑2-glycoprotein (LRG).37 Classification models have also been generated based on the selected discriminatory peaks between bladder cancer patients and healthy individuals.38 Preoperative and postoperative sera from two patients with bladder cancer were analysed using 2DE and revealed decreased expression of 68 protein spots in postoperative samples compared to the preoperative sera.35 Further evaluation of protein S100-A8 and protein S100-A9 levels was conducted in 77 tumour-tissue specimens using immunohisto chemistry. A scoring system was adopted based on the staining intensity and the percentage of positively stained cells. Significantly higher levels of protein S100-A8 were correlated with muscle invasion (P
