As current clinical and pathological variables have a limited ability to predict tumour recurrence, progression or patient survival,

Transcription

1 28 The Authors Mini-review Article PREDICTING SURVIVAL AFTER RADICAL CYSTECTOMY MARGULIS ET AL. BJUI BJU INTERNATIONAL Predicting survival after radical cystectomy for bladder cancer Vitaly Margulis, Yair Lotan*, Francesco Montorsi and Shahrokh F. Shariat* Departments of Urology, The University of Texas MD Anderson Cancer Center, *The University of Texas Southwestern Medical Center, Dallas, TX, USA, and Vita-Salute University, Milan, Italy Accepted for publication 8 January 28 Accurate prediction is essential for patient counselling, appropriate selection of treatments and determination of eligibility for clinical trials. In this review we assess the available determinants of oncological outcome after radical cystectomy (RC) for transitional cell carcinoma of the urinary bladder. We reviewed previous publications to provide guidelines in terms of criteria, limitations and clinical value of available tools for predicting patient outcome after RC. Our findings suggest that while individual surgical, patient and pathological features provide useful estimates of survival outcome, the inherent heterogeneity of tumour biology and patient characteristics leads to significant variation in outcome. By incorporating all relevant continuous predictive factors for individual patients, integrative predictive models, such as nomograms or artificial neural networks, provide more accurate predictions and generally surpass clinical experts at predicting outcomes. Nonetheless, there is a clear need for the development and validation of molecular biomarkers and their incorporation into multivariable predictive tools. Significant progress has been made in identifying important molecular markers of disease and the development of multifactorial tools for predicting the outcome after RC. KEYWORDS survival, radical cystectomy, urothelial carcinoma INTRODUCTION Bladder cancer is the fourth most frequent solid tumour among men and the seventh most frequent solid tumour among women, with >35 new cases diagnosed annually worldwide. In the USA, urothelial carcinoma (UC) of the bladder is estimated to affect 67 people, resulting in 13 7 deaths in 27 [1]. Most patients present with nonmuscle-invasive tumours, while 2 25% of patients have a muscle-invasive cancer as the first manifestation of their disease. Radical cystectomy (RC) is the treatment of choice for patients with muscle-invasive cancers and for selected patients with non-muscle-invasive cancers for whom intravesical therapy has failed or who are considered at high risk of progression to muscle-invasive disease. Despite advances in surgical techniques, improved understanding of the role of lymphadenectomy and progress in postoperative care, the 5-year diseasespecific survival after RC remains 5 6% [2,3]. Precise estimates of treatment success are essential for counselling patients about prognosis, appropriate selection of treatments and determination of eligibility for clinical trials. In particular, with the availability of effective systemic therapies, an accurate prediction of the presence of systemic micrometastatic disease and, consequently, the probability of progression after RC, is essential for selecting patients who might benefit from adjunctive chemotherapy [4]. Anatomical staging systems are the simplest examples of a prediction tool, categorizing the disease based on outcome. The American Joint Committee on Cancer TNM staging system has been validated and used universally to predict the risk of recurrence after RC [5]. While these staging systems have provided useful estimates of survival outcome, the inherent heterogeneity of tumour biology and patient characteristics leads to significant variation in outcomes within each staging and grading group. Moreover, current clinical and pathological staging systems for bladder cancer do not incorporate other important clinical, pathological and molecular markers of disease outcome. Furthermore, many patients with bladder cancer are elderly and have significant comorbid conditions, so that competing risks are important in evaluating outcomes. As current clinical and pathological variables have a limited ability to predict tumour recurrence, progression or patient survival, molecular markers have the potential to further stratify individuals according to their cancer-specific prognosis. Currently, several fairly large retrospective studies of archival material show the potential usefulness of several classical markers, and many smaller studies have suggested the usefulness of other identified candidate biomarkers [6]. Recently, prognostic modelling approaches, simultaneously considering many risk factors, have been used for assessing the risk in individual patients with colorectal, breast, prostate and renal cancers, and are easily adapted for clinical use [7]. More recently, significant advances have been made in the development of predictive tools that provide accurate risk estimates for patients with bladder cancer [8 11]. In this review we highlight important risk factors and available prognostic models for predicting outcomes after RC for bladder cancer. PATHOLOGICAL TUMOUR CHARACTERISTICS Tumour stage has been clearly established as one of the most important predictors of cancer-specific and overall survival after RC for bladder cancer [2,3]. The clinical TNM JOURNAL COMPILATION 28 BJU INTERNATIONAL 12, doi:1.1111/j x x 15

2 MARGULIS ET AL. staging system, used routinely for treatment decision-making and patient counselling, combines a pathological evaluation of the transurethral resection (TUR) specimen with findings from an examination under anaesthesia and preoperative radiographic imaging. Unfortunately, the inaccuracy of current imaging and under-sampling during TUR significantly limit the prognostic accuracy of clinical staging system before RC. Inaccuracies in clinical staging are highlighted in that up to 4% of patients are up-staged and roughly a quarter are down-staged after pathological assessment of the RC specimen [12]. The problem of under-staging has significant implications in counselling patients about neoadjuvant chemotherapy. After RC and lymph node dissection the pathological evaluation of the specimen provides a more accurate stratification of cancer-specific outcome. Many large singlecentre reports and several multi-institutional collaborations confirm the prognostic power of primary pt stage. The 5-year bladder cancer-specific survival (CSS) in patients with superficial (pt-pt1), muscle-invasive (pt2) and extravesical disease extending to perivesical fat (pt3) and adjacent organs (pt4) is reported as 8 9%, 5 7%, 3 45% and 2 35%, respectively [2,3]. The current 22 TNM staging system subclassifies pt2 tumours into two categories; cancer invading <5% of the depth of muscular propria (pt2a), and cancer invading >5% of the depth of muscularis propria [5]. Several studies have compared the clinical outcomes between pt2a and pt2b tumours, and have failed to find a prognostic difference between these tumour types [13,14]. Similarly, in several recent reports, the subdivision of pt3 tumours into pt3a (tumours with microscopic extravesical tumour extension) and pt3b (tumours with gross extravesical extension) categories, was not prognostically important [15]. The presence of metastatic lymph node deposits is the most important prognostic feature after RC, with associated 15 3% 5- year survival rates [16,17]. Recent evidence suggests that the number of positive nodes affects survival in node-positive patients. In one of the largest single-centre reports of patients treated with RC, the recurrence-free survival at 1 years for patients with eight or fewer positive lymph nodes was significantly higher than in those with eight or more positive lymph nodes (4% vs 1%, respectively) [2]. The same group of investigators reported that the total number of lymph nodes removed and identified by the pathologist affected survival in both nodenegative and -positive patients [18]. Consequently, a concept of lymph node density, which considers both the number of positive lymph nodes and the total number of nodes removed, was introduced and independently validated [16,19]. Tumour grading systems are designed to reflect the degree of tumour cell anaplasia. Even though there is no uniformly accepted grading system for bladder cancer, members of the WHO and the International Society of Urological Pathologists (ISUP) recently published the WHO/ISUP consensus classification of urothelial neoplasms of the urinary bladder. This system classifies bladder tumours into papillary urothelial neoplasms of low malignant potential, or papillary carcinomas of low or high grade [2]. While tumour grade is one of the most important determinants of tumour recurrence and progression after TUR of the tumour and/or intravesical immunotherapy, it has limited predictive power after RC [21]. Many studies clearly report a strong correlation between tumour grade and stage, but none has identified tumour grade as an independent predictor of bladder cancer-specific outcome after RC, largely because most patients undergoing RC have high-grade disease [2,3]. The current pathological TNM classification of bladder cancer relies on the depth of tissue invasion, but does not consider the size of the index tumour. However, in several recent reports, tumour size was identified as an independent predictor of bladder cancerspecific outcome [14,22,23]. In one study, the 1-year CSS was 94% for patients with pt2 tumours of 3 cm and only 68% for pt2 tumours of >3 cm (P <.1) [14]. While published proposed size threshold criteria have varied, there is a strong rationale to include tumour size as an important risk stratification variable for patients treated with RC. Lymphovascular invasion (LVI) has been identified as an independent predictor of cancer-specific outcome in many solid tumours. Tumour LVI has been identified in 3 5% of RC specimens, and positively correlates with adverse pathological features [24]. One large contemporary analysis of patients after RC confirmed the predictive power of LVI independent of lymph node status [25]. A multi-institutional study found that LVI is an independent predictor of bladder CSS in lymph node-negative, nonmetastatic patients treated with RC [25,26]. This is an important distinction, as patients with positive lymph nodes are usually recommended to have adjuvant chemotherapy regardless of LVI status. Knowing that patients with negative nodes and positive LVI are at higher risk of recurrence might improve the treatment in these patients. Despite ample evidence showing the independent predictive value of LVI after RC, the pathological criteria for identifying LVI have not been standardized. Moreover, none of the studies evaluated the prognostic significance of LVI prospectively. The most common histological subtype of bladder cancer in Western countries is UC, comprising >9% of cases [27]. Histological subtypes of non-uc of the bladder include mesenchymal and epithelial tumours of other histological types. Epithelial cancers include squamous cell, adenocarcinoma and small cell/neuroendocrine carcinoma. Nonepithelial subtypes include sarcoma, and less common tumours such as lymphoepithelioma-like carcinoma, phaeochromocytoma and melanoma [28]. The influence of non-uc subtype on clinical outcome has not been well defined due to the rarity of these tumours in Western countries. Several single-centre experiences and a large multi-institutional analysis failed to show a significant difference in bladder cancer-specific progression or death after RC when comparing squamous and transitional cell histology [27,29]. Conversely, non-transitional/non-squamous histology (i.e. adenocarcinoma, small cell carcinoma, carcinosarcoma, etc.) was identified as an independent predictor of bladder cancer-specific progression and mortality [27]. SURGICAL FACTORS The appropriate timing of RC after a diagnosis of bladder cancer has only recently been investigated. In one study, 16 patients were evaluated for the time from diagnosis to treatment. Those treated within 3 months of diagnosis had better recurrence-free, causespecific and overall survival than those who were treated later [3]. In another series of 247 patients, extravesical or node-positive disease was more prevalent in those who had 16 JOURNAL COMPILATION 28 BJU INTERNATIONAL

3 PREDICTING SURVIVAL AFTER RADICAL CYSTECTOMY RC >3 months after the initial diagnosis (84% vs 42%). Moreover, a delay in RC by 3 months was associated with a significantly lower rate of overall survival (35% vs 62%) [31]. In the same series, when adjusted for nodal status and pathological stage, the interval to treatment remained a statistically significant predictor of overall and CSS [31]; however, another study could not confirm these findings [32]. A positive surgical margin after RC is rare, but a strong independent predictor of local recurrence and bladder cancer-specific mortality. Complicating 2 1% of RCs, the finding of a positive surgical margin correlates with tumour stage, previous pelvic radiation, extent of lymph node dissection and surgeon experience [33]. A recent analysis by the Southwest Oncology Group 871 trial showed a 1% local recurrence rate and % 5-year CSS rate among 25 patients with a positive surgical margin after radical cystectomy [17]. A bilateral pelvic and iliac lymphadenectomy, including primary and secondary lymph node drainage of the bladder, is an integral part of RC and provides accurate staging and regional disease control [16]. A standard lymph node dissection removes all of the distal common iliac, external iliac, obturator and hypogastric nodes, and yields an average of 1 14 lymph nodes. An extended lymph node dissection removes lymph node tissue up to the aortic bifurcation, including the presacral nodes, and often yields 2 4 lymph nodes [34]. To date, accepted surgical standards for lymph node dissection or lymph node yield have not been validated. The notion that extended lymph node dissection minimizes the risk of leaving micrometastatic lymph node deposits in the retroperitoneum has been supported in several studies, which showed clearly that removing more lymph nodes during lymph node dissection for bladder cancer resulted in a diminished positive margin rate, decreased local recurrence and improved CSS [16,17,34 36]. The survival advantage offered by extended lymph node dissection has been reported in patients with lymph node-positive and -negative bladder cancer [17,37]. In one large series, the 5-year CSS rate for patients having cystectomy and no, standard or extended lymph node dissection was 33%, 46% and 6%, respectively [17]. The same authors reported a 44% 5-year survival rate for patients with fewer than 1 nodes removed, compared with 61% for patients with more than 1 nodes examined [17]. In summary, extended lymph node dissection allows for more accurate staging of the disease, and might also provide a therapeutic benefit. The minimum extension of the node dissection or an ideal minimum number of nodes can only be determined in a randomized trial, but it has been suggested that at least 15 2 lymph nodes should be obtained to assure staging accuracy and therapeutic benefit [16]. MOLECULAR MARKERS Mutations of cell-cycle regulatory genes are the genetic alterations most commonly found in human neoplasia, including UC of the bladder [38 41]. The tumour-suppressor p53 is known as the guardian of the genome because it can integrate many signals that control cell proliferation and apoptosis [42]. While most of the published material supports the view that p53 nuclear accumulation is predictive of the outcome in patients treated with RC [43 54], there is evidence for and against almost every aspect of the role of p53. The discrepancies between studies could be related to the choice of antibody, variability in the interpretation and stratification criteria, and inconsistency in specimen handling and technical procedures. Malats et al. [53] reviewed the published studies of p53 and noted that altered p53 expression was an independent prognostic factor for tumour recurrence in nine of 34 reports, for progression in 12 of 24, and for survival in 1 of 35. The retinoblastoma gene is the prototype tumour-suppressor gene. It encodes a nuclear protein (prb) that acts as a cell-cycle control checkpoint at the G1 phase [55]. A retrospective evaluation of prb expression in RC specimens showed significantly worse 5- year bladder CSS (33% vs 66%) in those patients with altered prb expression [56]. Several other investigators reported a stageindependent correlation of prb status with survival after RC for muscle-invasive bladder cancer [54,57]. The product of the p21 gene binds to and inhibits the activity of cyclin-dependent kinase-2 or -4 complexes, and thus functions as a regulator of cell-cycle progression at G1 [58]. In a large immunohistochemical study of p21 and p53 expression in 242 patients, multivariate analysis showed that p21 status was an independent predictor of tumour recurrence and survival after RC [47]. Moreover, patients with p53-altered/p21- negative tumours had a higher rate of disease relapse and worse CSS than those with p53- altered/p21-positive tumours [47]. Cyclin E1 is a pivotal regulatory molecule determining rates of cell-cycle transition from the G1 to S phase in the cyclin/cyclindependent kinase/prb pathway [59]. In a study of RC specimens from 226 consecutive patients with advanced TCC, low cyclin E1 expression was significantly associated with advanced pathological stage, LVI and lymph node metastases [6]. Moreover, in multivariate analyses, low cyclin E1 expression was significantly associated with bladder cancer-specific mortality. Ki-67 is a nuclear protein expressed by proliferating cells, and can be detected immunohistochemically. Nuclear Ki-67 antigen expression is a measure of the cell-growth fraction and hence biological aggressiveness of a malignancy [61]. In several large cystectomy series, Ki-67 overexpression was significantly associated with advanced pathological stage, higher tumour grade, LVI and metastases to lymph nodes [62 64]. Most importantly, authors established that Ki-67 expression was independently associated with both disease recurrence and bladder cancer-specific mortality [64]. Survivin is a member of the inhibitor of apoptosis gene family that has been found to control mitotic progression and induces changes in gene expression that are associated with tumour cell invasiveness [65]. In a recent analysis of 222 patients treated with RC, survivin expression was detected in 63% of the index bladder tumours, and correlated significantly with tumour grade, stage and lymph node status [66]. In a multivariate analysis adjusting for important prognostic features, altered survivin expression was an independent predictor of cancer-specific outcome [66]. Activation of caspases is a key event in apoptotic signalling and is required to execute the cell-death programme. Activated caspase- 3 constitutes an important downstream step in both the intrinsic and extrinsic apoptotic pathways [67]. In a recent study evaluating the combined effect of apoptotic markers on oncological outcomes after RC, altered JOURNAL COMPILATION 28 BJU INTERNATIONAL 17

4 MARGULIS ET AL. caspase-3 tissue expression was an independent predictor of bladder cancer recurrence and cancer-specific mortality [68]. These investigators also showed a cooperative effect of Bcl-2, caspase-3, p53 and survivin expression on outcomes after RC. When none of the markers was altered, recurrence-free and disease-specific survival were 75% and 89% at 7 years, respectively; when all four markers were altered, these values declined to 9% and 8% at 7 years, respectively [68]. Thrombospondin-1 (TSP-1) is an extracellular matrix glycoprotein that is a potent inhibitor of angiogenesis. A study evaluating TSP-1 expression in 163 RC specimens showed that TSP-1 expression was significantly associated with p53 expression status and microvessel density counts [69]. Moreover, decreased TSP- 1 expression was an independent predictor of disease recurrence and overall survival after stratifying for tumour stage, lymph node status, and histological grade [69]. Matrix metalloproteinases (MMPs) are a family of zinc-dependent enzymes that are considered to have an important cellular regulatory function by degrading the surrounding stroma, cytokines, and even membrane-bound receptors [7]. Abnormal MMP expression has been proposed to foster cancer development by stimulating cellular proliferation, promoting angiogenesis, or promoting invasion and metastases. In a study of 97 patients with advanced UC, the serum ratio of MMP-2 to MMP-2 inhibitor, determined by enzyme immunoassay, was a significant indicator of bladder cancer recurrence, independent of standard clinical and pathological variables [71]. Given the complexity of the molecular abnormalities associated with bladder cancer, it is improbable that a single marker can accurately segregate tumours of similar clinicopathological phenotypes into distinct prognostic categories. A marker might reflect the disruption of a biochemical pathway by a particular mechanism but not by another. Therefore, as previously shown, combinations of independent, complementary markers might provide a more accurate prediction of outcome than a single marker [43,48,49,72,73]. Karam et al. [68] found that p53, Bcl-2, caspase-3 and survivin show distinct association patterns with tumour stage and grade, LVI and carcinoma in situ (CIS). The number of simultaneously altered apoptosis markers was an important prognostic indicator for disease recurrence and bladder CSS in patients treated with RC. Changes in expression of markers were frequent in individuals with bladder cancer, with only 22 patients (1%) showing normal expression of all four markers. Alteration of fewer than four apoptosis markers did not provide independent prognostic information after controlling for the effects of standard pathological features; only when all four markers were altered was significance reached. This is explained by the fact that Bcl- 2, caspase-3, p53 and survivin have not only interrelated but also independent roles in the apoptotic pathway. In agreement with previous studies [48,49], Shariat et al. [5,54,74] confirmed that analysis of any combination of cell-cycle regulators (i.e. p53, prb, p21, p27, cyclin D1, and/or cyclin E1) provided additional prognostic information beyond that obtained from any single biomarker or combination of two or three biomarkers. This is rationale, because it investigates several pathways, including downstream effector pathways, rather than a single junction in a pathway. A higher total number of altered biomarkers was associated with a progressive, proportional increase in the risk of advanced pathological tumour stage, LVI, lymph node metastases, disease recurrence, and death from bladder cancer. In conclusion, several cell-cycle, apoptosis, angiogenesis and extracellular matrix modulator molecules have shown potential in providing important prognostic information related to cancer-specific outcomes after RC. The recent advances in biomarker discovery and validation are encouraging, but should be interpreted with caution. As a result of the lack of standardized technology to assess the accuracy of a particular prognostic marker, the lack of definitive criteria for evaluating test positivity, and absence of clearly defined endpoints and statistical methods, none of the molecular markers have successfully entered the clinical practice arena. OUTCOME PREDICTION MODELS NOMOGRAMS BASED ON STANDARD HISTOPATHOLOGICAL FEATURES Whereas the current TNM staging system continues to be the standard determinant of bladder CSS after RC, the heterogeneity of tumour biology and patient characteristics within each prognostic group results in significant variation of outcome within each staging category. Thus, incorporating other important prognostic variables discussed above, in addition to the TNM classification, might improve the prediction of outcomes in patients treated with RC. A nomogram is a graphical representation of a mathematical formula or algorithm that incorporates several predictors, modelled as continuous variables, to predict a particular endpoint [7]. By incorporating all relevant continuous predictive factors for individual patients, nomograms provide more accurate predictions than models based on risk grouping, and generally surpass clinical experts at predicting outcomes [75]. The inaccuracy of the clinical staging system before RC is well documented, but continues to be a major determinant governing therapeutic decision-making [12]. Consequently, the development of accurate preoperative risk-stratification models would allow the prediction of advanced disease and enable better selection of patients who would benefit from neoadjuvant systemic chemotherapy. To this end, nomograms for predicting before RC advanced pathological stage (pt3-4) and presence of lymph node metastases were developed from a multicentre cohort of 731 patients with available clinical and pathological staging data [1]. Using patient age, TUR stage, grade, and presence of CIS, these nomograms were 75.7% accurate in predicting advanced pathological stage (vs 71.4% using TUR stage alone) and 63.1% accurate in predicting lymph node metastases (vs 61.% using TUR T stage alone) [1]. While showing that adverse pathological features at RC can be predicted more accurately using several factors than with a single variable, nomograms before RC provide only a modest increase in accuracy. Differences in TUR technique, including the use of re-staging biopsies, and variability in the pathological evaluation, might have contributed to the suboptimal accuracy of these nomograms. Undoubtedly, incorporating other pathological prognostic markers, such as LVI, in addition to the molecular markers of disease, will enhance predictive accuracy of these nomograms. Several nomograms for use after RC have been developed to predict the natural history of surgically treated bladder cancer and to assist in the deciding on the use of adjuvant 18 JOURNAL COMPILATION 28 BJU INTERNATIONAL

5 PREDICTING SURVIVAL AFTER RADICAL CYSTECTOMY FIG. 1. A bladder CSS nomogram in 731 patients treated with RC and bilateral lymphadenectomy for UC of the bladder. Reprinted with permission from [9]. Points Pathologic 2 4 T stage LVI Pathologic 1 3 N stage 2 Neoadj. 1 Chemotherapy Adjuvant 1 radiotherapy Total Points 2Yrs.Surv.Prob. 5Yrs.Surv.Prob. 8Yrs.Surv.Prob. BLADDER CANCER-SPECIFIC SURVIVAL NOMOGRAM e-5 therapy after RC [8,9,76]. The Bladder Cancer Research Consortium used a multiinstitutional cohort of 731 consecutive patients treated with RC and bilateral lymphadenectomy for UC to determine the probabilities of recurrence, cancer-specific (Fig. 1) and all-cause mortality at 2, 5 and 8 years after RC [9,76]. To address each of the three outcomes, three separate nomograms were developed and internally validated. All three exceeded the accuracy of the TNM stage groupings, and of the individual predictors. All showed excellent performance characteristics, which virtually corresponded to ideal predictions. The recurrence nomogram (accuracy 78%) relied on pt and N stages, pathological grade, presence of LVI at RC, presence of CIS at RC, and the delivery of chemotherapy (either neoadjuvant or adjuvant or both) and/or radiation. The causespecific mortality nomogram showed 78% accuracy, vs 73% for all-cause mortality. The authors showed that their nomograms are significantly more accurate/discriminating than the TNM staging risk grouping, resolving FIG. 2. Postoperative nomogram that integrates the immunohistochemical status of five established cell-cycle regulatory biomarkers (p53, prb, p21, p27, and cyclin E1) with standard histopathological variables for predicting 1-, 2- and 5-year risk of disease recurrence in 191 patients with pta-3 NM UC of the bladder treated with RC and bilateral lymphadenectomy. Reprinted with permission from [77]. Points Path. T.stg LVI #Altered Markers Total Points 1(Yr) BCa-RFs 2(Yr) BCa-RFs 5(Yr) BCa-RFs T1;Tis;Ta T some of the heterogeneity of outcome prediction within each TNM-staging risk group. Moreover, the nomogram predictions were tailored to the risk posed by the characteristics of an individual s cancer, which is more relevant to the patient than are group-level probabilities. In the same year, the International Bladder Cancer Nomogram Consortium published a postoperative nomogram predicting the risk of recurrence at 5 years after RC and pelvic lymph node dissection [8]. The dataset developed for this study included >9 patients from 12 centres. Age, gender, grade, pathological stage, histological type, lymph node status, and time from diagnosis to surgery were significant contributing factors in the nomogram. The predictive accuracy of the nomogram (75%) was statistically better than TNM staging (68%) or standard pathological grouping models (62%). There are several limitations of nomogram approaches to patient risk stratification. First and foremost, all currently available predictive T tools in bladder cancer are not perfectly accurate. Because of difficulty with uniform data collection, inherent in multi-institutional collaborative efforts, important surgical variables, e.g. the timing of RC, margin status, extent of lymph node dissection and lymph node density, are not reflected in currently available predictive models. Furthermore, all available nomograms were derived and are applicable to centres of excellence for bladder surgery. Their applicability in the real world must be viewed with caution and certainly requires additional validation. Finally, any further improvement in the predictive accuracy of nomograms might require the integration of molecular prognosticators, as clinical and pathological prognosticators appear to have limited prognostic ability. NOMOGRAMS INCLUDING NOVEL BIOMARKERS The accuracy of current predictive tools is not yet perfect. Better modelling of the data, use of larger datasets, and more systematic and focused data collection (e.g. tighten the definition of symptom status) might help to improve the accuracy of the nomogram. The limited accuracy of current models is partly related to the heterogeneous biological behaviour of tumours with the same clinical and/or pathological features. The integration of novel biomarkers and/or data derived from imaging tools that are associated with the biological behaviour of bladder cancer might help to improve the accuracy of nomogram predictions. Despite numerous reports of promising new biomarkers in urological publications, only two studies have to date reported a statistically significant improvement in predictive accuracy when biomarkers were added to established predictors in the predictive tool setting [77 8]. Two smaller studies have added biomarkers to standard clinicopathological features, using artificial neural networks and neuro-fuzzy modelling [79,8]. Shariat et al. [77] showed that adding a panel of five well-established cell-cycle regulatory biomarkers (p53, prb, p21, p27, and cyclin E1) improved the predictive accuracy of competing-risk nomograms for predicting bladder cancer recurrence and survival after RC for patients with pta-pt3 node-negative tumours, by a clinically significant margin JOURNAL COMPILATION 28 BJU INTERNATIONAL 19

6 MARGULIS ET AL. (Fig. 2). Alterations in cell-cycle regulators were common, with 82% of patients having at least one altered biomarker, 2% three, and 16% four or five altered biomarkers. Patients with three or more altered biomarkers had a 4 1 times higher risk of bladder cancer recurrence and mortality after RC. Nomograms, such as these, that incorporate pathological and molecular information could form the basis for counselling patients about their risk of disease recurrence after surgery, and for designing clinical trials to test adjuvant treatment strategies in high-risk patients. CONCLUSIONS Precise estimation of cancer-specific outcomes after RC for UC is essential for counselling patients, selecting them for neoadjuvant and adjuvant systemic therapies, and determining their eligibility for clinical trials. The TNM staging system has been validated and used universally to predict the risk of recurrence after RC. However, anatomical staging systems for bladder cancer do not incorporate important clinical, surgical and molecular features known to affect the biology of UC. Not surprisingly, purely anatomical staging systems have limited ability to predict tumour recurrence, progression or patient survival. Recently, significant advances have been made in identifying important molecular markers of disease and in developing multifactorial tools for predicting the outcome after RC. Nomograms have provided patients and physicians with better individualized disease-related risk estimates that facilitate management-related decisions. Tools for accurately predicting patient outcomes are very important, as they affect the decision to use multimodal therapy. Reliance on anecdotal information has been faulty, both for clinicians and patients. Continuous, multivariable models such as nomograms are a highly appealing means of calculating accurate predictions with or without the use of a computer. Nomograms currently represent the most accurate and discriminating tools for predicting outcomes in patients with bladder cancer. CONFLICT OF INTEREST None declared. REFERENCES 1 Jemal A, Siegel R, Ward E, Murray T, Xu J, Thun MJ. Cancer statistics, 27. CA Cancer J Clin 27; 57: Stein JP, Lieskovsky G, Cote R et al. Radical cystectomy in the treatment of invasive bladder cancer: long-term results in 154 patients. J Clin Oncol 21; 19: Shariat SF, Karakiewicz PI, Palapattu GS et al. Outcomes of radical cystectomy for transitional cell carcinoma of the bladder: a contemporary series from the Bladder Cancer Research Consortium. J Urol 26; 176: Amiel GE, Lerner SP. Combining surgery and chemotherapy for invasive bladder cancer: current and future directions. Expert Rev Anticancer Ther 26; 6: Greene FL, American Joint Committee on Cancer, American Cancer Society. AJCC Cancer Staging Manual, 6th edn. New York: Springer-Verlag, 22: Habuchi T, Marberger M, Droller MJ et al. Prognostic markers for bladder cancer. International Consensus Panel on bladder tumor markers. Urology 25; 66: Kattan MW. Nomograms: an introduction. Semin Urol Oncol 22; 2: Bochner BH, Kattan MW, Vora KC. Postoperative nomogram predicting risk of recurrence after radical cystectomy for bladder cancer. J Clin Oncol 26; 24: Shariat SF, Karakiewicz PI, Palapattu GS et al. Nomograms provide improved accuracy for predicting survival after radical cystectomy. Clin Cancer Res 26; 12: Karakiewicz PI, Shariat SF, Palapattu GS et al. Precystectomy nomogram for prediction of advanced bladder cancer stage. Eur Urol 26; 5: Bassi P, Sacco E, De Marco V, Aragona M, Volpe A. Prognostic accuracy of an artificial neural network in patients undergoing radical cystectomy for bladder cancer: a comparison with logistic regression analysis. BJU Int 27; 99: Shariat SF, Palapattu GS, Karakiewicz PI et al. Discrepancy between clinical and pathologic stage: impact on prognosis after radical cystectomy. Eur Urol 27; 51: Pollack A, Zagars GK, Cole CJ, Dinney CP, Swanson DA, Grossman HB. The relationship of local control to distant metastasis in muscle invasive bladder cancer. J Urol 1995; 154: Cheng L, Weaver AL, Leibovich BC et al. Predicting the survival of bladder carcinoma patients treated with radical cystectomy. Cancer 2; 88: Quek ML, Stein JP, Clark PE et al. Microscopic and gross extravesical extension in pathological staging of bladder cancer. J Urol 24; 171: Stein JP. The role of lymphadenectomy in patients undergoing radical cystectomy for bladder cancer. Curr Oncol Rep 27; 9: Herr HW, Faulkner JR, Grossman HB et al. Surgical factors influence bladder cancer outcomes: a cooperative group report. J Clin Oncol 24; 22: Quek ML, Sanderson KM, Daneshmand S, Stein JP. The importance of an extended lymphadenectomy in the management of high-grade invasive bladder cancer. Expert Rev Anticancer Ther 24; 4: Herr HW. Superiority of ratio based lymph node staging for bladder cancer. J Urol 23; 169: Epstein JI, Amin MB, Reuter VR, Mostofi FK. The World Health Organization/International Society of Urological Pathology consensus classification of urothelial (transitional cell) neoplasms of the urinary bladder. Bladder Consensus Conference Committee. Am J Surg Pathol 1998; 22: Kirkali Z, Chan T, Manoharan M et al. Bladder cancer: epidemiology, staging and grading, and diagnosis. Urology 25; 66: Narayana AS, Loening SA, Slymen DJ, Culp DA. Bladder cancer: factors affecting survival. J Urol 1983; 13: Fung CY, Shipley WU, Young RH et al. Prognostic factors in invasive bladder carcinoma in a prospective trial of preoperative adjuvant chemotherapy and radiotherapy. J Clin Oncol 1991; 9: Algaba F. Lymphovascular invasion as a prognostic tool for advanced bladder cancer. Curr Opin Urol 26; 16: Quek ML, Stein JP, Nichols PW et al. Prognostic significance of 2 JOURNAL COMPILATION 28 BJU INTERNATIONAL

7 PREDICTING SURVIVAL AFTER RADICAL CYSTECTOMY lymphovascular invasion of bladder cancer treated with radical cystectomy. J Urol 25; 174: Lotan Y, Gupta A, Shariat SF et al. Lymphovascular invasion is independently associated with overall survival, cause-specific survival, and local and distant recurrence in patients with negative lymph nodes at radical cystectomy. J Clin Oncol 25; 23: Rogers CG, Palapattu GS, Shariat SF et al. Clinical outcomes following radical cystectomy for primary nontransitional cell carcinoma of the bladder compared to transitional cell carcinoma of the bladder. J Urol 26; 175: Eble JN. World Health Organization. Pathology and Genetics of Tumours of the Urinary System and Male Genital Organs. Lyon, Oxford: IARC Press, Oxford University Press, 24: Ghoneim MA, el-mekresh MM, el-baz MA, el-attar IA, Ashamallah A. Radical cystectomy for carcinoma of the bladder: critical evaluation of the results in 126 cases. J Urol 1997; 158: Lee CT, Madii R, Daignault S et al. Cystectomy delay more than 3 months from initial bladder cancer diagnosis results in decreased disease specific and overall survival. J Urol 26; 175: Sanchez-Ortiz RF, Huang WC, Mick R, Van Arsdalen KN, Wein AJ, Malkowicz SB. An interval longer than 12 weeks between the diagnosis of muscle invasion and cystectomy is associated with worse outcome in bladder carcinoma. J Urol 23; 169: Nielsen ME, Palapattu GS, Karakiewicz PI et al. A delay in radical cystectomy of >3 months is not associated with a worse clinical outcome. BJU Int 27; 1: Herr HW. Surgical factors in the treatment of superficial and invasive bladder cancer. Urol Clin North Am 25; 32: Poulsen AL, Horn T, Steven K. Radical cystectomy: extending the limits of pelvic lymph node dissection improves survival for patients with bladder cancer confined to the bladder wall. J Urol 1998; 16: Herr HW, Bochner BH, Dalbagni G, Donat SM, Reuter VE, Bajorin DF. Impact of the number of lymph nodes retrieved on outcome in patients with muscle invasive bladder cancer. J Urol 22; 167: Leissner J, Ghoneim MA, Abol-Enein H et al. Extended radical lymphadenectomy in patients with urothelial bladder cancer: results of a prospective multicenter study. J Urol 24; 171: Lerner SP, Skinner DG, Lieskovsky G et al. The rationale for en bloc pelvic lymph node dissection for bladder cancer patients with nodal metastases: longterm results. J Urol 1993; 149: Levine AJ, Momand J, Finlay CA. The p53 tumour suppressor gene. Nature 1991; 351: Hollstein M, Sidransky D, Vogelstein B, Harris CC. p53 mutations in human cancers. Science 1991; 253: Ruas M, Peters G. The p16ink4a/ CDKN2A tumor suppressor and its relatives. Biochim Biophys Acta 1998; 1378: F Esrig D, Spruck CH, Nichols PW et al. p53 nuclear protein accumulation correlates with mutations in the p53 gene, tumor grade, and stage in bladder cancer. Am J Pathol 1993; 143: Vogelstein B, Lane D, Levine AJ. Surfing the p53 network. Nature 2; 48: Cote RJ, Dunn MD, Chatterjee SJ et al. Elevated and absent prb expression is associated with bladder cancer progression and has cooperative effects with p53. Cancer Res 1998; 58: Sarkis AS, Dalbagni G, Cordon-Cardo C et al. Nuclear overexpression of p53 protein in transitional cell bladder carcinoma: a marker for disease progression. J Natl Cancer Inst 1993; 85: Sarkis AS, Bajorin DF, Reuter VE et al. Prognostic value of p53 nuclear overexpression in patients with invasive bladder cancer treated with neoadjuvant MVAC. J Clin Oncol 1995; 13: Serth J, Kuczyk MA, Bokemeyer C et al. p53 immunohistochemistry as an independent prognostic factor for superficial transitional cell carcinoma of the bladder. Br J Cancer 1995; 71: Stein JP, Ginsberg DA, Grossfeld GD et al. Effect of p21waf1/cip1 expression on tumor progression in bladder cancer. J Natl Cancer Inst 1998; 9: Shariat SF, Tokunaga H, Zhou J et al. p53, 21, prb, and p16 expression predict clinical outcome in cystectomy with bladder cancer. J Clin Oncol 24; 22: Chatterjee SJ, Datar R, Youssefzadeh D et al. Combined effects of p53, 21, and prb expression in the progression of bladder transitional cell carcinoma. J Clin Oncol 24; 22: Shariat SF, Ashfaq R, Sagalowsky AI, Lotan Y. Correlation of cyclin D1 and E1 expression with bladder cancer presence, invasion, progression, and metastasis. Hum Pathol 26; 37: Esrig D, Elmajian D, Groshen S et al. Accumulation of nuclear p53 and tumor progression in bladder cancer. N Engl J Med 1994; 331: Cote RJ, Esrig D, Groshen S, Jones PA, Skinner DG. p53 and treatment of bladder cancer. Nature 1997; 385: Malats N, Bustos A, Nascimento CM et al. P53 as a prognostic marker for bladder cancer: a meta-analysis and review. Lancet Oncol 25; 6: Shariat SF, Zlotta AR, Ashfaq R, Sagalowsky AI, Lotan Y. Cooperative effect of cell-cycle regulators expression on bladder cancer development and biologic aggressiveness. Mod Pathol 27; 2: Benedict WF, Xu HJ, Hu SX, Takahashi R. Role of the retinoblastoma gene in the initiation and progression of human cancer. J Clin Invest 199; 85: Cordon-Cardo C, Wartinger D, Petrylak D et al. Altered expression of the retinoblastoma gene product: prognostic indicator in bladder cancer. J Natl Cancer Inst 1992; 84: Logothetis CJ, Xu HJ, Ro JY et al. Altered expression of retinoblastoma protein and known prognostic variables in locally advanced bladder cancer. J Natl Cancer Inst 1992; 84: Harada K, Ogden GR. An overview of the cell cycle arrest protein, p21 (WAF1). Oral Oncol 2; 36: Yasmeen A, Berdel WE, Serve H, Muller-Tidow C. E- and A-type cyclins as markers for cancer diagnosis and prognosis. Expert Rev Mol Diagn 23; 3: Kamai T, Takagi K, Asami H, Ito Y, Oshima H, Yoshida KI. Decreasing of p27 (Kip1) and cyclin E protein levels is associated with progression from superficial into invasive bladder cancer. Br J Cancer 21; 84: Cattoretti G, Becker MH, Key G et al. Monoclonal antibodies against JOURNAL COMPILATION 28 BJU INTERNATIONAL 21

8 MARGULIS ET AL. recombinant parts of the Ki-67 antigen (MIB 1 and MIB 3) detect proliferating cells in microwave-processed formalinfixed paraffin sections. J Pathol 1992; 168: Popov Z, Hoznek A, Colombel M et al. The prognostic value of p53 nuclear overexpression and MIB-1 as a proliferative marker in transitional cell carcinoma of the bladder. Cancer 1997; 8: Suwa Y, Takano Y, Iki M, Asakura T, Noguchi S, Masuda M. Prognostic significance of Ki-67 expression in transitional cell bladder carcinoma after radical cystectomy. Pathol Res Pract 1997; 193: Margulis V, Shariat SF, Ashfaq R, Sagalowsky AI, Lotan Y. Ki-67 is an independent predictor of bladder cancer outcome in patients treated with radical cystectomy for organ-confined disease. Clin Cancer Res 26; 12: Li F. Role of survivin and its splice variants in tumorigenesis. Br J Cancer 25; 92: Shariat SF, Ashfaq R, Karakiewicz PI, Saeedi O, Sagalowsky AI, Lotan Y. Survivin expression is associated with bladder cancer presence, stage, progression, and mortality. Cancer 27; 19: Fernandes-Alnemri T, Litwack G, Alnemri ES. CPP32, a novel human apoptotic protein with homology to Caenorhabditis elegans cell death protein Ced-3 and mammalian interleukin-1 beta-converting enzyme. J Biol Chem 1994; 269: Karam JA, Lotan Y, Karakiewicz PI et al. Use of combined apoptosis biomarkers for prediction of bladder cancer recurrence and mortality after radical cystectomy. Lancet Oncol 27; 8: Ioachim E, Michael MC, Salmas M et al. Thrombospondin-1 expression in urothelial carcinoma: prognostic significance and association with p53 alterations, tumour angiogenesis and extracellular matrix components. BMC Cancer 26; 6: 14 7 Chirco R, Liu XW, Jung KK, Kim HR. Novel functions of TIMPs in cell signaling. Cancer Metastasis Rev 26; 25: Gohji K, Fujimoto N, Ohkawa J, Fujii A, Nakajima M. Imbalance between serum matrix metalloproteinase-2 and its inhibitor as a predictor of recurrence of urothelial cancer. Br J Cancer 1998; 77: Cordon-Cardo C, Zhang ZF, Dalbagni G et al. Cooperative effects of p53 and prb alterations in primary superficial bladder tumors. Cancer Res 1997; 57: Shariat SF, Kim J, Raptidis G, Ayala GE, Lerner SP. Association of p53 and p21 expression with clinical outcome in patients with carcinoma in situ of the urinary bladder. Urology 23; 61: Shariat SF, Ashfaq R, Sagalowsky AI, Lotan Y. Predictive value of cell cycle biomarkers in nonmuscle invasive bladder transitional cell carcinoma. J Urol 27; 177: Specht MC, Kattan MW, Gonen M, Fey J, Van Zee KJ. Predicting nonsentinel node status after positive sentinel lymph biopsy for breast cancer: clinicians versus nomogram. Ann Surg Oncol 25; 12: Karakiewicz PI, Shariat SF, Palapattu GS et al. Nomogram for predicting disease recurrence after radical cystectomy for transitional cell carcinoma of the bladder. J Urol 26; 176: Shariat SF, Karakiewicz PI, Ashfaq R et al. Multiple biomarkers improve prediction of bladder cancer recurrence and mortality in patients undergoing cystectomy. Cancer 28; 112: Shariat SF, Zippe C, Ludecke G et al. Nomograms including nuclear matrix protein 22 for prediction of disease recurrence and progression in patients with Ta, T1 or CIS transitional cell carcinoma of the bladder. J Urol 25; 173: Qureshi KN, Naguib RN, Hamdy FC, Neal DE, Mellon JK. Neural network analysis of clinicopathological and molecular markers in bladder cancer. J Urol 2; 163: Catto JW, Linkens DA, Abbod MF et al. Artificial intelligence in predicting bladder cancer outcome: a comparison of neurofuzzy modeling and artificial neural networks. Clin Cancer Res 23; 9: Correspondence: Vitaly Margulis, Department of Urology, The University of Texas MD Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 773, USA. vmargu@hotmail.com Abbreviations: RC, radical cystectomy; UC, urothelial carcinoma; CSS, cancer-specific survival; TUR, transurethral resection; ISUP, International Society of Urological Pathologists; LVI, lymphovascular invasion; TSP-1, thrombospondin-1; MMP, matrix metalloproteinase; CIS, carcinoma in situ. 22 JOURNAL COMPILATION 28 BJU INTERNATIONAL