Review – Bladder Cancer

Nomograms for Bladder Cancer

By: Shahrokh F. Shariata lowast , Vitaly Margulisb, Yair Lotana, Francesco Montorsic and Pierre I. Karakiewiczd

European Urology, Volume 54 Issue 1, July 2008, Pages 41-53

Published online: 01 July 2008

Keywords: Bladder cancer, Nomogram, Prediction, Prognosis, Risk

Abstract Full Text Full Text PDF (826 KB)



Patients with bladder cancer face a variable risk of recurrence based on their clinical characteristics and the biology of their disease. Physicians need tools to accurately estimate the risk of recurrence and cancer-specific mortality to recommend individualized therapy and to design appropriate clinical trials.


A MEDLINE literature search was performed on bladder cancer nomograms from January 1966 to July 2007. We recorded input variables, prediction form, number of patients used to develop the prediction tools, the outcome being predicted, prediction tool-specific features, predictive accuracy, and whether validation was performed. Each prediction tool was classified into patient clinical disease state and the outcome being predicted.


The literature search generated 11 published prediction tools that may be applied to patients in various clinical stages of bladder cancer. Of the 11 prediction tools, 8 have undergone validation. The following considerations need to be applied when designing and judging predictive models: predictive accuracy (internal and external validation), calibration, generalizability (reproducibility and transportability), and level of complexity, with the intent of determining whether the new model offers advantages relative to available alternatives. Studies comparing decision tools show that nomograms outperform other methodologies such as risk grouping.


Nomograms provide the most accurate individualized risk estimations that facilitate management decisions. However, current nomograms still need to be refined. Potential advances may include the incorporation of biomarkers, validation in larger patient cohorts, and prospective data acquisition.

Take Home Message

This article provides guidelines in the process of decision aid selection for bladder cancer patients. We provide an overview of recent bladder cancer predictive tools organized by clinical state (non–muscle-invasive versus muscle-invasive bladder cancer). We recorded predictor variables, the outcome of interest, the number of patients used to develop the tools, tool-specific features, predictive accuracy estimates, and whether internal or external validation was performed.

Keywords: Bladder cancer, Nomogram, Prediction, Prognosis, Risk.

1. Introduction

Carcinoma of the urinary bladder, the fourth most common cancer in men and the ninth most common cancer in women, results in significant morbidity and mortality [1]. There will be an estimated 67,160 new cases and 13,750 deaths in men and women in the United States in 2007, the vast majority of which are urothelial carcinoma (UC). At initial diagnosis, about 70–80% of patients have bladder cancers that are confined to the epithelium or subepithelial connective tissue. These cancers can be managed with endoscopic resection and intravesical therapy. The recurrence rate for these tumors is 50–70% and 10–15% progress to muscle invasion over a 5-yr period [2].

The remaining 20–30% of patients present with muscle-invasive cancer at initial diagnosis. Of this population, 50% have distant metastasis within 2 yr, and 60% die within 5 yr despite treatment [3] and [4]. Progression to measurable metastatic disease occurs on average 1–2 yr after radical cystectomy and is fatal in the majority of patient despite a high initial response rate to systemic chemotherapy [3] and [4].

Accurate estimates of the likelihood of treatment success, complications, and long-term morbidity are essential for patient counseling and informed decision-making. A well-informed patient will be more likely to be compliant with treatment suggestions and studies suggest that a lack of patient involvement is a major risk factor for regret of treatment choice [5]. Finally, accurate risk estimates can help identify homogeneous high-risk patient groups to assist with clinical trial design.

Traditionally, physician judgment has formed the basis for risk estimation, patient counseling, and decision-making. However, clinicians’ estimates might be biased due to subjective and objective confounders that exist at all stages of the prediction process [6], [7], [8], and [9]. Clinicians do not recall all cases equally; certain cases can stand out and exert an unsuitably large influence when predicting future outcomes. Clinicians might be inconsistent when processing their memory and tend to resort to heuristics (‘rules of thumb’) when processing becomes difficul [10]. Finally without resorting to computers, clinicians might find it difficult to integrate the growing numbers of predictive variables that are important in clinical decision-making [11] and [12]. To circumvent these limitations and to obtain the most accurate and reliable predictions, researchers have developed predictive/prognostic tools based on statistical models. Recently, predictive/prognostic nomograms have been introduced to complement the standard modeling techniques with the ability to predict various risks for individual patients. In general, these predictive models have performed as well as or better than clinical judgment when predicting probabilities of outcome [12]. That said, physician input is obviously essential and crucial for the measurement of variables that are used in the prediction process, as well as in the interpretation and application of model-derived outcome predictions in clinical practice.

Decision aids used for bladder cancer consist of the “Kattan-type” nomograms, risk groupings, artificial neural networks (ANNs), and probability tables. Within the last 5 yr, the field of predictive modeling has exploded. In this review we catalogue the current nomograms available for bladder cancer. We describe the patient populations to which they apply and the outcomes predicted, and record their individual characteristics. Moreover, this review may serve as an initial step toward a comprehensive reference guide for physicians to locate published nomograms that apply to the clinical decision in question.

2. “Kattan-type” nomograms

Various distinct statistical methodologies have broadly been described as “nomograms.” According to the strict definition, a nomogram represents a graphical calculation device that can be based on any type of function, such as logistic regression or Cox regression models [10] and [13]. The nomogram usually incorporates continuous or categorical variables. The effect of the variables on the outcome of interest is represented in the format of axes and risk points are attributed according to the prognostic/predictive importance of the variable of interest. For example, the nomogram in Fig. 1[14] assigns to each pathologic stage a unique point value that represents its prognostic significance. The ‘Total Points’ axis is used to estimate the combined effect of all predictor variables on the probability of the outcome. The nomogram format is unique because it allows combining the input of several continuously coded variables or that of several categorically coded variables. This format distinguishes nomograms from look-up tables or decision trees, where continuously coded variables cannot be processed and where data availability limits the degree of stratification to avoid empty cells or dead-end branches.


Fig. 1

Bladder cancer-specific survival nomogram in 731 patients treated with radical cystectomy and bilateral lymphadenectomy for urothelial carcinoma of the bladder. Instructions for nomogram use: Locate patient values at each axis. Draw a vertical line to the “Point” axis to determine how many points are attributed for each variable value. Sum the points for all variables. Locate the sum on the “Total Points” line. Draw a vertical line towards the “2Yrs.Surv.Prob.,” “5Yrs.Surv.Prob.,” and “8Yrs.Surv.Prob.” axes to determine, respectively, the 2- yr, 5-yr, and 8-yr survival probabilities. Reprinted with permission of Shariat et al. Clin Cancer Res 2006;12:6663 [14].

Nomograms are designed to extract the maximum amount of information from data with the goal of providing the most accurate predictions. The accuracy, or discriminant properties of prognostic nomograms, those that predict either recurrence or mortality rates are measured using the concordance index (or c-index), rather than the receiver–operator characteristic curve area. The area under the curve (AUC) requires binary outcomes (presence or absence of cancer) and is reserved for binary logistic regression models. The c-index represents an adaptation of the AUC for censored data and is necessary when time-to-event data are used [15]. The AUC quantifies the ability to discriminate between those with or without the outcome of interest. The c-index quantifies the ability of the nomogram to identify between two randomly chosen patients the one who relapses first. As for the AUC, a c-index of 0.5 represents no discriminating ability and a value of 1.0 represents perfect discrimination.

The available bladder cancer nomograms have been adapted for use on personal digital assistants and personal computers to facilitate their integration into daily clinical practice and research ( and

3. Evaluating predictive tools

Several considerations apply when designing and judging predictive models (Table 1):

Table 1

Available predictive models in bladder cancer

ReferencePrediction formPatient populationOutcomeNo. of patientsVariablesAccuracyValidation
Parmar et al [32]Risk groupingNon–muscle- invasive UCRecurrence-free survival919Number of tumors and cystoscopy at 3 moNot reportedNot performed
Millan-Rodriguez et al [33] and [42]Risk groupingNon–muscle- invasive UCRecurrence-free survival, progression-free survival, and all-cause survival1529Number of tumors, tumor size, T category, CIS, grade, intravesical BCGNot reportedNot performed
Shariat et al [30]Probability nomogramNon–muscle-invasive UCRecurrence-free survival and progression-free survival2681Age, gender, urine cytology, dichotomized NMP22® level (and institution)84% for recurrence of any UCInternal
87% for recurrence of high-grade UC or T1 and higher stage
86% for recurrence of stage ≥T2 UC
Sylvester et al [34]Look-up tableNon–muscle- invasive UCRecurrence-free survival and progression-free survival2596Number of tumors, tumor size, prior recurrence rate, T category, CIS, and gradeNot reportedNot performed
Quershi et al [40]Artificial neural networkTa T1Recurrence within 6 mo56EGFR, c-erbB2, p53, stage, grade, tumor size, number of tumors, gender, smoking status, histology of mucosal biopsies, CIS, metaplasia, architecture, location75%Internal
Ta T1Progression-free survival10580%
T2–T41-yr cancer-specific survival4082%
Catto et al [41]Neuro-fuzzy modelingTa–T4Recurrence-free survival109p53, mismatch repair proteins, stage, grade, age, smoking status, previous cancer88–95%Internal
Karakiewicz et al [36]Probability nomogramRadical cystectomyCystectomy T and N stage731Age, TUR stage, TUR grade, CIS76% for T stageInternal
63% for N stage
Karakiewicz et al [37]Probability nomogramRadical cystectomy2-, 5-, and 8-yr recurrence-free survival731Age, T stage, N stage, grade, LVI, CIS, adjuvant radiotherapy, adjuvant chemotherapy, neoadjuvant chemotherapy78%Internal
Shariat et al [14]Probability nomogramRadical cystectomy2-, 5-, 8-yr all-cause and bladder cancer-specific survival731Age, T stage, N stage, grade, LVI, adjuvant radiotherapy, adjuvant chemotherapy, neoadjuvant chemotherapy79% for all-cause survivalInternal
73% for bladder cancer-specific survival
Bochner et al [38]Probability nomogramRadical cystectomy5-yr recurrence-free survival9,064Age, gender, T stage, N stage, grade, histology, time from diagnosis to surgery75%Internal
Bassi et al [39]Artificial neural networkRadical cystectomy5-yr all-cause survival369Age, gender, T stage, N stage, LVI, grade, concomitant prostate cancer, history of upper tract UC76%Internal

UC = urothelial carcinoma; CIS = carcinoma in situ; BCG = bacillus Calmette-Guérin; EGFR = epidermal growth factor receptor; TUR = transurethral resection; LVI = lymphovascular invasion.

3.1. Predictive accuracy

Accuracy represents the most important consideration. Predictive accuracy should ideally be confirmed in an external cohort, which represents the best gold standard for validation. Internal validation represents an alternative and may consist of bootstrapping, leave-one-out, and split sample validation. Of internal validity tests, bootstrapping-derived accuracy estimates are the closest to external validity-derived estimates [10], [16], [17], [18], [19], and [20]. The receiver operating characteristic (ROC) AUC as well as the c-index indicate the discriminatory properties of a model and quantify overall accuracy. No model is perfect and generally accepted accuracy ranges are 70–80%. This implies that the risk of interest will be misclassified in between 20 and 30 of 100 patients. An accuracy gain of 2%, related to the consideration of a novel marker, implies that 2 of 100 additional patients will be correctly classified. Although such an increase may appear trivial, the effect of a 2% gain is substantially more relevant on a larger scale, when thousands of patients are considered. A small accuracy gain is also important when the nomogram is used for risk stratification within a clinical protocol. A more accurate nomogram will contribute to better distribution between study arms. Finally, a seemingly trivial gain is always important to patients who deserve the most accurate prognosis.

3.2. Calibration

A model with an overall 80% ability to discriminate between the presence and absence of the outcome of interest may perform well in one range of predicted probabilities. Conversely, it may perform substantially worse in another range. Therefore, the assessment of model's performance characteristics, termed calibration, is crucial. Calibration plots show the relationship between predicted and observed probabilities of the outcome of interest. The calibration plot of a model that predicts perfectly well shows a 45° line, which indicates perfect agreement between predicted and observed rates. It is imperative for clinicians to know the performance characteristics of the model they routinely use in clinical practice because some models may demonstrate substantially worse performance in external samples [16], [18], [19], [20], and [21].

3.3. Generalizability

General applicability of the model is important because patient and model characteristics may vary and may undermine the accuracy and the performance characteristics (calibration) of the model. The performance of a predictive instrument can decline from the development set to an external validation set, in part because of change in the predictability of risk factors. Nomogram development depends heavily on its development cohort and models can only be as good as the development cohort data are. Thus, prior to using a tool, the clinician should ensure that the nomogram is applicable to the patient [16], [18], [19], [20], and [21]. Generalizability limitations may be related to differences in disease characteristics, differences in population characteristics, or due to stage or grade migration. For example, a model that is specific to transitional cell carcinoma of the urinary bladder cannot be applied to patients with squamous cell variants. Inclusion or exclusion criteria might also affect generalizability. For example, several models were developed using single-center databases [22]. Such models may not be invariably applicable to community practice [23]. Finally, stage migration may affect the natural history of the disease. Patients treated with radical cystectomy for bladder cancer 20 yr ago might not show the same disease characteristics as their contemporary counterparts

3.4. Level of complexity

The level of complexity represents an important consideration. Excessively complex models are clearly impractical in busy clinical practices. For example, models that require pathologic characteristics that are not routinely recorded or molecular information that is not accessible to most clinicians will not be implemented into clinical practice. Similarly, models that require computational infrastructure might pose problems with their applicability. For example, ANNs can accurately predict several outcomes, but the use of ANNs might be restricted due to lack of access to ANN code or lack of ANN-specific computer infrastructure.

3.5. Head-to-head comparison

Finally, when judging a new tool, one should examine its predictive accuracy, validity, and performance characteristics relative to established models, with the intent of determining whether the new model offers advantages relative to available alternatives [14], [17], [19], [20], [24], [25], [26], and [27]. Objective comparisons should focus on accuracy and performance characteristics. Subsequently, accuracy and performance characteristics should be compared in the light of generalizability and complexity. A simple model that relies on two variables might be better than a competing model with 10 variables, even if accuracy is slightly lower.

4. Limitations of nomograms

Besides obvious limitations related to accuracy, performance characteristics, generalizability, and the level of complexity, the most common potential additional limitations of currently available predictive and prognostic tools may be classified in one or several of the following categories.

4.1. Study selection criteria

Specific model criteria, such as inclusion and exclusion criteria, do not allow the use of models for patients with different characteristics or who have been exposed to different treatment modalities. For example, if a model development cohort excluded patients treated with neoadjuvant chemotherapy therapy, then predictions cannot be made for such patients.

4.2. Change over time of the predictive value of model ingredients

Changes in practice patterns represent important phenomena that affect cancer control rates. External validation in contemporary cohorts is necessary to ensure temporal validity.

4.2.1. Conditional probabilities

It would be beneficial to adjust for disease-free interval since surgery [28] and [29]. Unfortunately, no such tool exists for bladder cancer. Absence of adjustment for disease-free interval presents the clinician with an excessively somber estimate of cancer control over time. Expectedly, the latter improves with increasing disease-free interval.

4.2.2. Suboptimal predictive accuracy

No prediction model developed to date is perfect. This might be due to lack of consideration of all potential predictive risk factors and from the inability to assemble all known prognostic factors optimally. Bladder cancers with the same histopathologic features have a heterogeneous biologic behavior. Therefore, there is a need for novel biomarkers and imaging tools that are associated with the biologic behavior of bladder cancer to enhance the predictive accuracy of current tools [30] and [31].

5. Currently available prediction tools

The above discussion is meant to provide guidelines in the process of decision aid selection. Herein, we provide an overview of recent bladder cancer predictive tools organized by clinical state (non–muscle-invasive vs. muscle-invasive bladder cancer). We recorded predictor variables, the outcome of interest, the number of patients used to develop the tools, tool-specific features, predictive accuracy estimates, and whether internal or external validation has been performed.

5.1. Prediction of disease recurrence and progression in patients with history of non–muscle-invasive bladder cancer

Previous efforts to improve risk stratification have led to the development of risk grouping models in which several reliable predictors were combined. The British Medical Research Council focused on establishing risk groupings to predict the development of recurrence of Ta and T1 tumors. These investigators demonstrated that disease status at the 3-mo cystoscopy and tumor multifocality were the most important predictors of subsequent recurrence and thus incorporated into their risk schema [32].

Millan-Rodriguez et al assessed predictors of recurrence, progression, and mortality among 1529 patients with superficial bladder cancer and showed that patients can be stratified into different risk groups, based on multifocality, tumor size, intravesical bacillus Calmette-Guérin (BCG) therapy, and presence of carcinoma in situ (CIS) [33]. Tumor grade was the most powerful predictor of disease progression and disease-specific mortality.

The European Association of Urology (EAU) guidelines stratify patients into different risk groups: low risk (single lesion, Ta, grade 1, and ≤3 cm in diameter) versus intermediate risk (Ta–T1, grade 1–2, multifocal, >3 cm) versus high risk (any T1, grade 3, multifocal or highly recurrent, CIS).

The first nomogram in bladder cancer was published in 2005 [30]. In a multi-institutional collaboration among Austria, Canada, Egypt, Germany, Japan, Spain, Sweden, Switzerland, and the United States, the authors developed nomograms that estimated the risk of disease recurrence and progression in 2681 patients with Ta, T1, and Tis bladder cancer (Fig. 2). All subjects had previous histologically confirmed non–muscle-invasive bladder cancer. All patients provided voided urine samples for cytologic and NMP22® analyses before undergoing cystoscopy. Patients with suspicious cystoscopy or cytology results were further investigated with transurethral biopsies. Of 2681 patients (mean age 65 yr; 75% men, 25% women) enrolled in the study, 956 had recurrent UC. Histologic data were available for 898 of these patients: 24% had grade 1, 43% grade 2, and 33% grade 3 tumors, and 45% had stage Ta, 32% stage T1 or CIS, and 23% stage ≥T2 tumors. Patient age, urine cytology status, and urinary NMP22® level were associated with all three end points in univariable and multivariable analyses (p < 0.001). The predictive accuracy of a model based on patient age, gender, and urine cytology was significantly increased for all three end points when NMP22® level was included as a variable. The predictive accuracy of each model varied considerably by institution. The most accurate model incorporated patient age, gender, cytology, dichotomized NMP22® level (cut-off of 10 U/ml), and the institution of origin. Its accuracy after bootstrap validation was 0.842 for recurrence of any UC; 0.869 for recurrence of grade 3, stage Ta or T1 UC, or CIS; and 0.858 for recurrence of stage ≥T2 UC (Fig. 2). A similar model (based on age, gender, cytology, and dichotomized NMP22®) demonstrated near-perfect performance for prediction of any UC recurrence or recurrence of grade 3, stage Ta or T1 UC, or CIS. These nomograms can be accessed free-of-charge at


Fig. 2

Nomograms for (A) recurence of any transitional cell carcinoma; (B) recurrence of grade 3 Ta or T1 or of carcinoma in situ (CIS); (C) recurrence of ≥T2 stage transitional cell carcinoma in 2681 patients who underwent office cystoscopy for detection of bladder cancer recurrence after treatment of stage Ta, T1, or CIS urothelial carcinoma of the urinary bladder. Reprinted with permission of Shariat et al. J Urol 2005;173:1518 [30].

The principal application of the NMP22-based recurrence nomogram could provide a means for individualizing the cystoscopy follow-up in patients with Ta, T1, or CIS bladder UC. The nomogram could help define the ideal timing of repeat cystoscopy. During follow-up, instead of undergoing repeat cystoscopic evaluations, the indication for cystoscopy could be defined according to individual recurrence risk. According to this risk, patients could be either reassessed at a later date or undergo cystoscopy. The authors, however, acknowledged that the decision to repeat cystoscopy must be made on a case-by-case basis. The procedure should be undertaken according to the predicted probability for bladder cancer, patient clinical history, and the minimal risk that the patient is willing to accept before deciding to undergo another cystoscopy session and the likelihood of finding a tumor-free bladder. Moreover, this model has several limitations. It does not consider several established predictor variables, such as previous pathologic grade and stage, number, and pattern of previous recurrences, time since the original diagnosis, and prior use of intravesical therapy. Moreover, the performance of the nomograms varied significantly among the institutions, emphasizing the need for external validation of all tools.

Sylvester et al recently developed a look-up table based on data from 2596 patients with stage Ta T1 UC who were entered in seven European Organization for Research and Treatment of Cancer (EORTC) trials (Fig. 3) [34]. Data from BCG trials were not considered. The EORTC model allows urologists to calculate the probabilities of recurrence and progression over a period of 1–5 yr based on a patient's clinical and pathologic tumor characteristics at the time of transurethral resection (TUR) such as the prior recurrence rate, number and size of tumors, the stage and grade of the tumors, and the presence of concomitant CIS. The probability of recurrence varies from 15% at 1 yr in patients with a good prognosis to >75% at 5 yr for those with the worst prognosis. Similarly, the probability of progression varies from 0.2% at 1 yr to 45% at 5 ys. Certain subgroups of patients were found to have a particularly poor prognosis. For example, the probability of progression in T1G3 patients with concomitant CIS was 29% at 1 yr and 74% at 5 yr. In internal validation, recurrence predictions were 66% accurate at 1 and 5 yr; progression predictions were 74% and 75% accurate, respectively. The software has been developed to allow urologists to easily calculate a patient's probabilities of recurrence and progression, either at the initial diagnosis or on recurrence. The software for this model is available at


Fig. 3

European Organization for Research and Treatment of Cancer scoring system and risk tables for stage Ta T1 bladder cancer. The risk tables allow estimation of the probability of recurrence and progression in patients with stage Ta T1 bladder cancer based on number of tumors, tumor size, prior recurrence rate, T category, concomitant carcinoma in situ, and grade.

An important limitation of the Sylvester et al modeling is that the development data sets predate the common use of BCG maintenance therapy, which is known to reduce/delay recurrence and progression rates, respectively. The authors are currently testing the validity of their tool in patients on BCG maintenance therapy (R. Sylvester, pers. comm.). Moreover the data predate the standard second TUR of bladder tumor (TURBT) in high-risk patients. In addition, approximately 20% of the patients did not receive any additional intravesical treatment, and <10% received an immediate instillation.

Whereas the nomogram from Shariat et al takes into account age, gender, pre-cystoscopy cytology, and NMP22® to predict recurrence and progression during the follow up of patients with a previous history of superficial bladder cancer [30], the EORTC tables predict a patient's future short- and long-term probabilities of recurrence and progression at the time of the initial diagnosis or at the time of a recurrence based on the clinical and pathologic characteristics [34]. Thus, the two nomograms serve different and complimentary purposes.

5.2. Preoperative prediction of pathologic features at radical cystectomy

Inaccuracy of the pre-cystectomy clinical staging system is well documented, but it continues to be major determinant governing therapeutic decision-making [35]. Consequently, development of accurate preoperative risk-stratification models would allow prediction of advanced disease and enable better selection of patients who would benefit from neoadjuvant systemic chemotherapy. To this end, pre-cystectomy nomograms for the prediction of advanced pathologic stage (pT3–4) and presence of lymph node metastases were developed from a multicenter cohort of 731 patients with available clinical and pathologic staging data [36]. When patient age, TUR stage, grade, and presence of CIS were integrated within the nomogram, 75.7% accuracy was recorded in predicting advance pathologic stage (pT3–4) versus 71.4% when TUR stage alone was used. The nomogram was 63.1% accurate in predicting lymph node metastases when TUR stage and grade were used versus 61.0% using TUR T stage alone. The various components of these nomograms can be accessed at

The pre-cystectomy nomograms provide only a modest increase in accuracy. Nonetheless, they demonstrate that the combined use of clinical and pathologic variables, which cannot always be integrated within look-up tables, results in more accurate predictions than the use of a single variable. Several variables may have contributed to the suboptimal accuracy of these nomograms. These may include differences in TUR technique, nonstandardized use of restaging biopsies, and variability in the pathologic evaluation. Possibly, the integration of other pathologic prognostic markers, such as lymphovascular invasion (LVI), in addition to molecular markers of disease, might enhance predictive accuracy of pre-cystectomy nomograms. However, the limited ability to predict nodal metastases indicates that the accurate pretreatment prediction of this outcome represents a challenge in urologic oncology.

5.3. Postoperative prediction of disease recurrence and survival after radical cystectomy

Several post-cystectomy nomograms have been developed to predict the natural history of surgically treated bladder cancer and to assist in the decision-making process regarding the use of adjuvant therapy after radical cystectomy [14], [37], and [38]. The Bladder Cancer Research Consortium (BCRC) used a multi-institutional cohort of 731 consecutive patients treated with radical cystectomy and bilateral lymphadenectomy for UC to determine the probabilities of recurrence and cancer-specific and all-cause mortality at 2, 5, and 8 yr after surgery [14] and [37]. To address each of the three outcomes, three separate nomograms were developed and internally validated. All three exceeded the accuracy of the American Joint Committee on Cancer (AJCC) stage groupings, as well as 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, pathologic grade, presence of LVI at cystectomy, presence of CIS at cystectomy, as well as the delivery of chemotherapy (either neoadjuvant or adjuvant or both) or radiation or both. The cause-specific mortality nomogram showed 78% accuracy versus 73% for all-cause mortality.

The authors demonstrated that their nomograms are significantly more accurate/discriminating than the AJCC staging risk grouping, resolving some of the heterogeneity of outcome prediction within each AJCC 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 (IBCNC) published a postoperative nomogram predicting the risk of recurrence at 5 yr following radical cystectomy and pelvic lymph node dissection [38]. The data set developed for this study included >9000 patients from 12 centers. Age, gender, grade, pathologic stage, histologic 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 superior to either AJCC TNM staging (68%) or standard pathologic grouping models (62%).

The recurrence nomogram of the IBCNC is more generalizable than that of the BCRC and can be applied to patients with histologic variants other than UC. For example, most patients originated from Mansoura University in Egypt. Conversely, the BCRC nomogram is best suited for patients from the United States and may be exclusively applied to patients with UC. Despite these limitations, the BCRC nomogram offers a 4% accuracy advantage (78% vs. 74%) compared to the IBCNC nomogram. This may be due to a higher homogeneity of the population in the BCRC study. In addition, the BCRC offers a suite of nomograms predicting all-cause and cancer-specific survival in addition to tumor recurrence. Furthermore, their models provide 2-, 5- and 8-yr predictions, whereas the IBCNC nomogram provides only a 5-yr prediction. This feature allows better identification of early bladder cancer recurrences, which represent the most aggressive type of recurrences.

Several limitations of nomogram approaches to patient risk stratification should be noted. First, and foremost, all currently available predictive tools in bladder cancer are not perfectly accurate. Because of difficulty with uniform data collection, inherent in multi-institutional collaborative efforts, important surgical parameters, such as timing of cystectomy, 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 centers 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 predictive accuracy of nomograms may require the integration of molecular prognosticators because clinical and pathologic prognosticators appear to have limited prognostic ability.

As an alternative to nomogram-based modeling, Bassi et al developed an ANN using gender, age at surgery, LVI, pT, pN, grade, presence of concomitant prostatic adenocarcinomam and history of upper tract urothelial tumors as input variables for prediction of 5-yr all-cause survival after cystectomy [39]. In a single-institution cohort of 369 patients, the prognostic accuracy of the ANN (76%; based on 12 variables) was slightly superior to the logistic regression model that was based on only two statistically significant variables (75%; stage and grade) [39]. Unfortunately, the comparison of the accuracy of both models was performed on the same population that served for model development, which undermines the validity of such comparison.

5.4. Nomograms including novel biomarkers

The accuracy of current predictive tools is not yet perfect. Better modeling of the data, use of larger data sets, and more systematic and focused data collection (eg, tighten the definition of symptom status) may help improve the accuracy of nomograms. The limited accuracy of current models is partially related to the heterogeneous biologic behavior of tumors with the same clinical or pathologic features. The integration of novel biomarkers or data derived from imaging tools that are associated with the biologic behavior of bladder cancer may help improve the accuracy of nomogram predictions.

Over the past two decades, the molecular dissection of cancer has increased our understanding of the pathways that are altered in neoplastic cells. Protein expression profiling of bladder cancer offers an alternative means to distinguish aggressive tumor biology and may improve the accuracy of outcome prediction. In addition, the emergence of new therapeutic approaches for bladder cancer cannot flourish without a set of markers to serve as prognosticators or therapeutic targets. However, despite numerous reports of promising new biomarkers in the urologic literature, only two studies have to date demonstrated a statistically significant improvement in predictive accuracy when biomarkers were added to established predictors in the predictive tool setting [30], [31], [40], and [41]. Two smaller studies have added biomarkers to standard clinicopathologic features using ANNs and neuro-fuzzy modeling [40] and [41].

Shariat et al demonstrated that the addition of 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 prediction of bladder cancer recurrence and survival following cystectomy for patients with pTa–pT3 node-negative tumors by a clinically significant margin (Fig. 4) [31]. Alterations in cell cycle regulators were common with 82% of patients exhibiting at least one altered biomarker, 20% exhibiting three altered biomarkers, and 16% exhibiting four or five altered biomarkers. Patients with three or more altered biomarkers had a 4- to 10-times elevated risk of bladder cancer recurrence and mortality after radical cystectomy. Nomograms, such as these, that incorporate pathologic and molecular information could form the basis for counseling patients regarding their risk of disease recurrence following surgery and for designing clinical trials to test adjuvant treatment strategies in high-risk patients.


Fig. 4

Postoperative nomogram that integrates the immunohistochemical status of five established cell cycle regulatory biomarkers (p53, pRB, p21, p27, and cyclin E1) with standard histopathologic variables for predicting 1-, 2- and 5-yr risk of disease recurrence in 191 patients with pTa–3 N0 M0 urothelial carcinoma of the bladder treated with radical cystectomy and bilateral lymphadenectomy. Instructions for physicians: Locate the patient's T stage on the sex axis. Draw a straight line up to the points axis to determine how many points toward recurrence the patient should receive. Repeat this process for each of the remaining axes, drawing a straight line each time to the points axis. Sum the points received for each predictive variable and locate this number on the total points axis. Draw a straight line down from the total points to one of the recurrence-free prediction (RFS) axes for the patient's specific risk of remaining free from recurrence for 1, 2, and 5 yr. Reprinted with permission of Shariat et al. Cancer. In press [31].

6. Conclusions

Tools for accurate prediction of patient outcomes are extremely important because they affect the decision to use multimodality therapy. Reliance on anecdotal information has proven to be faulty both for clinicians and patients. Due to the plethora of different decision tools, it is important to understand the mechanisms by which they work and the advantages and disadvantages of each. Our review demonstrates that nomograms offer a more accurate and practical approach than ANNs and risk groupings. The addition of molecular markers to nomograms may add to standard pathologic features. Use of nomograms in decision-making may improve in individualized patient care.

At a minimum, patients with bladder cancer need to be involved in the decision regarding management of their disease. They should know what their options are and what the consequences can be. Ideally, patients would make their own treatment decisions. At the core of any patient involvement is accurate prediction of consequences and, essentially, a spreadsheet of these predictions tailored to the individual. This spreadsheet represents informed consent for any medical decision. Providing this should reduce the likelihood of regret of treatment choice, particularly when complications arise.

Continuous, multivariable models such as nomograms are a highly appealing means of calculating accurate predictions with or without the use of a computer. Many nomograms have been constructed for patients with bladder cancer. Nomograms currently represent the most accurate and discriminating tools for predicting outcomes in patients with bladder cancer. When faced with the difficult decision of choosing among the treatment options for each clinical stage of bladder cancer, the nomograms provide patients with accurate estimates of outcomes. Equipped with this information, the patient is more likely to be confident in the treatment decision and less likely to experience regret in the future. However, it should be emphasized that nomogram predictions must be interpreted as such; they do not make treatment recommendations or act as a surrogate for physician–patient interactions nor do they provide definitive information on symptomatic disease progression or complications associated with treatments.

Unfortunately, no prospective randomized studies are available to clearly demonstrate that the use of nomograms improves patient care. However, as long as randomized trials cannot be used to dictate clinical decision-making, as is the case for testis or breast cancer, nomograms represent the best option for making the most informed decisions for clinicians and patients alike. Although most nomogram predictions are sufficiently accurate, they require the expert interpretation of physicians.

Many more nomograms, as well as improvements to existing nomograms, are needed. For example, none of the nomograms predicts with perfect accuracy. Novel biomarkers, larger data sets, better data collection methods, and more sophisticated modeling procedures are needed to improve predictive accuracy. In addition, better accuracy might be accomplished by modeling physician- or hospital-specific data for patients being treated by that physician or at that hospital. Finally, nomograms that predict the likelihood of metastatic progression, cancer-specific mortality, and long-term quality of life are likely to have great utility for the patient and physician when exploring treatment alternatives. In summary, nomograms have empowered patients and physicians in their fight against bladder cancer by providing superior individualized disease-related risk estimations that facilitate management-related decisions.

Conflicts of interest

The authors have nothing to disclose.


  • [1] A. Jemal, R. Siegel, E. Ward, T. Murray, J. Xu, M.J. Thun. Cancer statistics, 2007. CA Cancer J Clin. 2007;57:43-66 Crossref
  • [2] J.A. Witjes, K. Hendricksen. Intravesical pharmacotherapy for non–muscle-invasive bladder cancer: a critical analysis of currently available drugs, treatment schedules, and long-term results. Eur Urol. 2008;53:45-52 Crossref
  • [3] J.P. Stein, G.D. Grossfeld, D.A. Ginsberg, et al. Prognostic markers in bladder cancer: a contemporary review of the literature. J Urol. 1998;160:645-659
  • [4] S.F. Shariat, P.I. Karakiewicz, G.S. Palapattu, et al. Outcomes of radical cystectomy for transitional cell carcinoma of the bladder: a contemporary series from the bladder cancer research consortium. J Urol. 2006;176:2414-2422 Crossref
  • [5] B.J. Miles, B. Giesler, M.W. Kattan. Recall and attitudes in patients with prostate cancer. Urology. 1999;53:169-174 Crossref
  • [6] A.S. Elstein. Heuristics and biases: selected errors in clinical reasoning. Acad Med. 1999;74:791-794
  • [7] I. Vlaev, N. Chater. Game relativity: how context influences strategic decision making. J Exp Psychol Learn Mem Cogn. 2006;32:131-149 Crossref
  • [8] M. Kattan. Expert systems in medicine. N.J. Smelser, P.B. Baltes (Eds.) International encyclopedia of the social and behavioral sciences (Pergamon, Oxford, United Kingdom, 2001) 5135-5139 Crossref
  • [9] R.M. Hogarth, N. Karelaia. Heuristic and linear models of judgment: matching rules and environments. Psychol Rev. 2007;114:733-758 Crossref
  • [10] M.W. Kattan. Nomograms. Introduction. Semin Urol Oncol. 2002;20:79-81
  • [11] F. Rabbani, A.M. Stapleton, M.W. Kattan, T.M. Wheeler, P.T. Scardino. Factors predicting recovery of erections after radical prostatectomy. J Urol. 2000;164:1929-1934
  • [12] P.L. Ross, C. Gerigk, M. Gonen, et al. Comparisons of nomograms and urologists’ predictions in prostate cancer. Semin Urol Oncol. 2002;20:82-88
  • [13] M.W. Kattan, J.A. Eastham, A.M. Stapleton, T.M. Wheeler, P.T. Scardino. A preoperative nomogram for disease recurrence following radical prostatectomy for prostate cancer. J Natl Cancer Inst. 1998;90:766-771 Crossref
  • [14] S.F. Shariat, P.I. Karakiewicz, G.S. Palapattu, et al. Nomograms provide improved accuracy for predicting survival after radical cystectomy. Clin Cancer Res. 2006;12:6663-6676 Crossref
  • [15] C.B. Begg, L.D. Cramer, E.S. Venkatraman, J. Rosai. Comparing tumour staging and grading systems: a case study and a review of the issues, using thymoma as a model. Stat Med. 2000;19:1997-2014
  • [16] Bradley E. a. T., R.J. Monographs on statistics and applied probability: an introduction to the bootstrap. Champman and Hall/CRC, 1993. p. 275–281.
  • [17] M.W. Kattan. Comparison of Cox regression with other methods for determining prediction models and nomograms. J Urol. 2003;170:S6-S9 discussion S10
  • [18] E.W. Steyerberg, S.E. Bleeker, H.A. Moll, D.E. Grobbee, K.G. Moons. Internal and external validation of predictive models: a simulation study of bias and precision in small samples. J Clin Epidemiol. 2003;56:441-447 Crossref
  • [19] E.W. Steyerberg, F.E. Harrell Jr, G.J. Borsboom, M.J. Eijkemans, Y. Vergouwe, J.D. Habbema. Internal validation of predictive models: efficiency of some procedures for logistic regression analysis. J Clin Epidemiol. 2001;54:774-781 Crossref
  • [20] E.W. Steyerberg, F.E. Harrell Jr, P.H. Goodman. Neural networks, logistic regression, and calibration. Med Decis Making. 1998;18:349-350 Crossref
  • [21] E.W. Steyerberg, M.J. Roobol, M.W. Kattan, T.H. van der Kwast, H.J. de Koning, F.H. Schroder. Prediction of indolent prostate cancer: validation and updating of a prognostic nomogram. J Urol. 2007;177:107-112 discussion 112 Crossref
  • [22] A.W. Partin, M.W. Kattan, E.N. Subong, et al. Combination of prostate-specific antigen, clinical stage, and Gleason score to predict pathological stage of localized prostate cancer. A multi-institutional update [see comments] [published erratum appears in JAMA 1997;278:118].. JAMA. 1997;277:1445-1451 Crossref
  • [23] D.F. Penson, G.D. Grossfeld, Y.P. Li, J.M. Henning, D.P. Lubeck, P.R. Carroll. How well does the Partin nomogram predict pathological stage after radical prostatectomy in a community based population? Results of the cancer of the prostate strategic urological research endeavor. J Urol. 2002;167:1653-1657 discussion 1658
  • [24] M.W. Kattan. Nomograms are superior to staging and risk grouping systems for identifying high-risk patients: preoperative application in prostate cancer. Curr Opin Urol. 2003;13:111-116 Crossref
  • [25] M.W. Kattan, S.F. Shariat, B. Andrews, et al. The addition of interleukin-6 soluble receptor and transforming growth factor beta1 improves a preoperative nomogram for predicting biochemical progression in patients with clinically localized prostate cancer. J Clin Oncol. 2003;21:3573-3579 Crossref
  • [26] F.K. Chun, M. Graefen, A. Briganti, et al. Initial biopsy outcome prediction—head-to-head comparison of a logistic regression-based nomogram versus artificial neural network. Eur Urol. 2007;51:1236-1243 (discussion 41–3) Crossref
  • [27] F.K. Chun, P.I. Karakiewicz, A. Briganti, et al. A critical appraisal of logistic regression-based nomograms, artificial neural networks, classification and regression-tree models, look-up tables and risk-group stratification models for prostate cancer. BJU Int. 2007;99:794-800 Crossref
  • [28] A.J. Stephenson, P.T. Scardino, J.A. Eastham, et al. Postoperative nomogram predicting the 10-year probability of prostate cancer recurrence after radical prostatectomy. J Clin Oncol. 2005;23:7005-7012 Crossref
  • [29] A.J. Stephenson, P.T. Scardino, J.A. Eastham, et al. Preoperative nomogram predicting the 10-year probability of prostate cancer recurrence after radical prostatectomy. J Natl Cancer Inst. 2006;98:715-717 Crossref
  • [30] S.F. Shariat, C. Zippe, G. Ludecke, 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. 2005;173:1518-1525 Crossref
  • [31] S.F. Shariat, P.I. Karakiewicz, R. Ashfaq, et al. Multiple biomarkers improve prediction of bladder cancer recurrence and mortality in patients undergoing cystectomy. Cancer. 2008;112:315-325 Crossref
  • [32] M.K. Parmar, L.S. Freedman, T.B. Hargreave, D.A. Tolley. Prognostic factors for recurrence and followup policies in the treatment of superficial bladder cancer: report from the British Medical Research Council Subgroup on Superficial Bladder Cancer (Urological Cancer Working Party). J Urol. 1989;142:284-288
  • [33] F. Millan-Rodriguez, G. Chechile-Toniolo, J. Salvador-Bayarri, J. Palou, F. Algaba, J. Vicente-Rodriguez. Primary superficial bladder cancer risk groups according to progression, mortality and recurrence. J Urol. 2000;164:680-684 Crossref
  • [34] R.J. Sylvester, A.P.M. van der Meijden, W. Oosterlinck, et al. Predicting recurrence and progression in individual patients with stage Ta T1 bladder cancer using EORTC risk tables: a combined analysis of 2596 patients from seven EORTC trials. Eur Urol. 2006;49:466-477 (discussion 75–7) Crossref
  • [35] S.F. Shariat, G.S. Palapattu, P.I. Karakiewicz, et al. Discrepancy between clinical and pathologic stage: impact on prognosis after radical cystectomy. Eur Urol. 2007;51:137-151 Crossref
  • [36] P.I. Karakiewicz, S.F. Shariat, G.S. Palapattu, et al. Precystectomy nomogram for prediction of advanced bladder cancer stage. Eur Urol. 2006;50:1254-1262 (discussion 1261–2) Crossref
  • [37] P.I. Karakiewicz, S.F. Shariat, G.S. Palapattu, et al. Nomogram for predicting disease recurrence after radical cystectomy for transitional cell carcinoma of the bladder. J Urol. 2006;176:1354-1361 discussion 1361–2
  • [38] B.H. Bochner, M.W. Kattan, K.C. Vora. Postoperative nomogram predicting risk of recurrence after radical cystectomy for bladder cancer. J Clin Oncol. 2006;24:3967-3972
  • [39] P. Bassi, E. Sacco, V. De Marco, M. Aragona, A. Volpe. Prognostic accuracy of an artificial neural network in patients undergoing radical cystectomy for bladder cancer: a comparison with logistic regression analysis. BJU Int. 2007;99:1007-1012 Crossref
  • [40] K.N. Qureshi, R.N. Naguib, F.C. Hamdy, D.E. Neal, J.K. Mellon. Neural network analysis of clinicopathological and molecular markers in bladder cancer. J Urol. 2000;163:630-633
  • [41] J.W. Catto, D.A. Linkens, M.F. Abbod, et al. Artificial intelligence in predicting bladder cancer outcome: a comparison of neuro-fuzzy modeling and artificial neural networks. Clin Cancer Res. 2003;9:4172-4177
  • [42] F. Millan-Rodriguez, G. Chechile-Toniolo, J. Salvador-Bayarri, J. Palou, J. Vicente-Rodriguez. Multivariate analysis of the prognostic factors of primary superficial bladder cancer. J Urol. 2000;163:73-78 Crossref


a Department of Urology, The University of Texas Southwestern Medical Center, Dallas, TX, USA

b Department of Urology, The University of Texas M.D. Anderson Cancer Center, Houston, TX, USA

c Department of Urology, Vita-Salute University, Milan, Italy

d Cancer Prognostics and Health Outcomes Unit, University of Montreal, Montreal, QC Canada

Corresponding author. Department of Urology, The University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX 75390-9110, USA. Tel. +1 469 363 8500; Fax: +1 214 648 8786.

Place a comment

Your comment *

max length: 5000