Abstract
Prostate cancer (PCa) is a heterogeneous disease with different disease states. Clinical judgment has significant biases. Nomograms are graphical calculating tools that use several clinical variables to determine a specific clinical outcome. Nomograms as prediction tools use multiple variables and continuous variables continuously, and their characteristics include patient populations, clinical end points, accuracy, and whether internal and/or external validations were performed. Numerous nomograms are available for most PCa stages, and their availability and accuracy make them powerful tools for the improvement of clinical prediction and important aids in personalised medicine for both patients and physicians.
Keywords
1. Introduction
Prostate cancer (PCa) is the most common malignancy in European and North American men. According to estimates in 2006, 345 900 (20.3%) new cases of PCa were diagnosed in Europe out of 1.7 million new cancer cases in males [
[1]
]. In the United States, 189 000 newly diagnosed PCa cases occurred in 2007 [[2]
]. Since the use of serum prostate-specific antigen (PSA) as a useful marker for PCa screening, a dramatic change has occurred in the pattern of patients who are diagnosed with PCa. Despite the controversies regarding the survival benefit of PCa screening by PSA [3
, 4
], its effect on downstaging at diagnosis is clear, and the association between PSA screening and PCa mortality has been established in recent studies [[5]
]. Most patients are being diagnosed with clinically confined disease, and the rate of metastatic disease at presentation has dramatically decreased: 20% (1973–1979), 16% (1985–1989), and 5% (1995–2001).PCa has unique features that complicate the decision-making process. The natural history of the disease is prolonged because of the relatively indolent biology and lead time of diagnosis. Different clinical variables exist, such as clinical stage, PSA level, and pathology parameters (Gleason score, number of involved cores, and percentage of involved tissue). Different treatments methods are available—surgery, radiation therapy (RT), minimally invasive procedures, and medical and active surveillance—and the efficacy of local therapy differs in correlation with the clinical variables.
2. Evidence acquisition
Clinical judgment is the basis for patient evaluation and for decision making in medicine; however, it is prone to several biases. Clinicians do not recall all cases equally, and the prediction of outcome is frequently chosen according to the desired one rather than to the one with the higher probability. The individual impact of the different clinical variables on outcome prediction varies, and the clinician's ability to evaluate each variable is limited. The decision-making process for an individual with localised PCa needs must consider additional information beyond the cancer data, including the patient's age, comorbidities, life expectancy, and preferences regarding potential toxicity. Traditionally, the solution for the above-mentioned limitations was the creation of risk groups. A specific clinical variable was used to classify patients into these risk groups, and then additional variables were used for further classification. The patient could have high-grade cancer according to his prostate biopsy, and further classification would be based on his PSA and clinical stage. The use of such models is easy and rapid; however, their accuracy is not optimal. The reliance on a single variable also creates heterogeneous groups regarding therapy outcomes within a specific risk group.
Despite the data from randomised, controlled trials on large patient populations, the individual patient needs an answer for his personal situation. Do all patients with positive surgical margins benefit from adjuvant RT following radical prostatectomy (RP)? Should chemotherapy be used for all patients with hormone-refractory prostate cancer (HRPC)? It is clear that additional information is needed beyond the message from the available randomised, controlled trials that evaluated the specific clinical question.
An alternative clinical tool for improvement of clinical judgment and to aid in personalised medicine is the nomogram. Nomograms are graphical calculating tools, usually with two dimensions, that use several clinical variables to provide an answer to a specific question. Nomograms are usually based on regression analysis and achieve a better accuracy prediction compared to risk groups as a result of their ability to use multiple variables and continuous variables continuously. Another advantage is that nomograms are based on multiple variables according to their impact on the predictive model rather than screening by initial univariate analysis.
For PCa patients, an important change in the estimation of prognosis emerged a decade ago. Kattan and colleagues published an important paper evaluating the outcome following local therapy for PCa according to pretherapy clinical data [
[6]
]. Since then, many nomograms have been introduced for PCa that include different disease states. In order to consider the usefulness of a specific nomogram, several issues must be reviewed besides the clinical question to be answered, including the nomogram's accuracy, its correlation between the predicted and observed risk, and whether a similar prediction was reached using an external dataset different from the one used for the creation of the model. The accuracy of a nomogram is indicated by its concordance index, ranging from 0.5 to 1. The closer the index is to 1, the better its predictive value. Because the model is based on the specific data from which it is created, the performance of external validation (the use of a different dataset for the model testing) is important.3. Evidence synthesis
Recently, an inventory of PCa predictive tools classified according to their end points was published. One hundred eleven different tools were published [
[7]
]. The advantage of such a review is the ability of the individual physician (or patient) to verify the available tools for the specific end point and to evaluate it according to the inherent characteristics. Examples of nomograms classified by their patient population and end point are provided in Table 1, Table 2, Table 3, Table 4 (adapted from Shariat et al [[7]
]). An additional aspect of the nomogram's use beyond its performance is it availability and complexity. Some of the current models are available for personal use, such as for personal computers or online. As a result of the incorporation into a digital format, the process of adding patient clinical data and getting the specific outcome is short, and is often completed within a few seconds. Because of their advantages, several nomograms (Kattan's preoperative nomogram [[6]
] and Briganti's lymph node nomogram [[9]
]) have been recommended by the recent PCa guidelines of the European Association of Urology as a useful tool for patient consultation [[8]
].Table 1Nomograms predicting biochemical recurrence following radical prostatectomy according to preoperative variables*
Patient population | End point | Author | Number | Accuracy | Validation |
---|---|---|---|---|---|
RP | BCR | Kattan et al, 1998 [6] | 983 | 74 (internal) | Internal and external |
RP | BCR | Stephenson et al, 2006 [11] | 1978 | 76–79 | Internal and external |
RP = radical prostatectomy; BCR = biochemical recurrence.
*Adapted from Shariat et al
[7]
.Table 2Nomograms predicting biochemical recurrence following radical prostatectomy according to postoperative variables*
Patient population | End point | Author | Number | Accuracy | Validation |
---|---|---|---|---|---|
RP | BCR | Kattan et al, 1999 [12] | 996 | 89 (internal) | Internal and external |
RP | BCR | Stephenson et al, 2005 [13] | 1881 | 78–86 | Internal and external |
RP = radical prostatectomy; BCR = biochemical recurrence.
*Adapted from Shariat et al
[7]
.Table 3Nomograms predicting biochemical recurrence following radiation therapy according to pretreatment variables*
Patient population | End point | Author | Number | Accuracy | Validation |
---|---|---|---|---|---|
External RT | BCR | Kattan et al, 2000 [14] | 1042 | 73 | Internal |
Brachytherapy | BCR | Kattan et al, 2001 [15] | 920 | 61 | Internal and external |
RT = radiation therapy; BCR = biochemical recurrence.
*Adapted from Shariat et al
[7]
.Table 4Nomograms predicting metastasis progression or survival*
Patient population | End point | Author | Number | Accuracy | Validation |
---|---|---|---|---|---|
External RT | Metastasis progression | Kattan et al, 2003 [16] | 1677 | 81 | Internal and external |
Rising PSA following RP or RT | Metastasis progression | Slovin et al, 2005 [17] | 148 | 69 | None |
Rising PSA following RP | Metastasis progression | Dotan et al, 2005 [18] | 248 | 93 | Internal |
Androgen-independent PCa | PCa-specific death | Svatek et al, 2006 [19] | 129 | 81 | Internal |
Progressive metastasis after castration | OS | Smaletz et al, 2002 [20] | 433 | 71 | Internal and external |
ADT following RP | PCa-specific death | Porter et al, 2007 [21] | 66 | 66 | Internal |
Metastatic HRPC | OS | Halabi et al, 2003 [22] | 1101 | 68 | Internal and external |
RT = radiation therapy; PSA = prostate-specific antigen; RP = radical prostatectomy; PCa = prostate cancer; OS = overall survival; ADT = androgen-deprivation therapy; HRPC = hormone-refractory prostate cancer.
*Adapted from Shariat et al
[7]
.Despite the above-mentioned advantages of nomograms as prognostic prediction tools, several disadvantages should be considered. A nomogram is based on the dataset from which it is built. Most of the data are retrospective and have a selection bias, that is, a selection of patients for a specific local therapy. Consequently, the selection can be associated with a specific outcome. An example of the influence of dataset characteristics on the performance of the prediction tool is the prediction of positive lymph nodes during RP. Different datasets showed that the number of positive lymph nodes at RP ranges between 1% and 11% [
[9]
]. However, the probability of positive lymph nodes depends not only on the tumour and patient characteristics, but also on the surgical template, the surgeon's performance, and the number of removed lymph nodes during surgery. Therefore, the prediction of the probability of positive lymph nodes according to a nomogram based on a limited lymph node dissection's template might lead to inaccurate prediction. The variables used for the creation of a model are those available at a specific time. Additional variables, such as new biomarkers or imaging modality results, might add predictive impact [[10]
]. In addition, most nomograms are based on datasets from academic centres, either from Europe or from the United States. Their outcomes might differ from outcomes achieved in community hospitals and different geographic areas, because the nomograms do not take into consideration important variables such as the quality of the provided therapy.4. Conclusions
Despite their limitations, nomograms created from large and high-quality datasets can be transformed into accurate predictions and improve the process of clinical decision making. Nomograms are useful tools for improving clinical prediction and personalised medicine for both patients and physicians.
Conflicts of interest
The authors have nothing to disclose.
Funding support
None.
References
- Estimates of the cancer incidence and mortality in Europe in 2006.Ann Oncol. 2007; 18: 581-592
- Cancer statistics, 2008.CA Cancer J Clin. 2008; 58: 71-96
- Screening and prostate-cancer mortality in a randomized European study.N Engl J Med. 2009; 360: 1320-1328
- Mortality results from a randomized prostate-cancer screening trial.N Engl J Med. 2009; 360: 1310-1319
- Interpreting trends in prostate cancer incidence and mortality in the five Nordic countries.J Natl Cancer Inst. 2007; 99: 1881-1887
- A preoperative nomogram for disease recurrence following radical prostatectomy for prostate cancer.J Natl Cancer Inst. 1998; 90: 766-771
- An updated catalog of prostate cancer predictive tools.Cancer. 2008; 113: 3075-3099
- EAU guidelines on prostate cancer.Eur Urol. 2008; 53: 68-80
- Pelvic lymph node dissection in prostate cancer.Eur Urol. 2009; 55: 1251-1265
- 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
- Preoperative nomogram predicting the 10-year probability of prostate cancer recurrence after radical prostatectomy.J Natl Cancer Inst. 2006; 98: 715-717
- Postoperative nomogram for disease recurrence after radical prostatectomy for prostate cancer.J Clin Oncol. 1999; 17: 1499-1507
- Postoperative nomogram predicting the 10-year probability of prostate cancer recurrence after radical prostatectomy.J Clin Oncol. 2005; 23: 7005-7012
- Pretreatment nomogram for predicting the outcome of three-dimensional conformal radiotherapy in prostate cancer.J Clin Oncol. 2000; 18: 3352-3359
- Pretreatment nomogram for predicting freedom from recurrence after permanent prostate brachytherapy in prostate cancer.Urology. 2001; 58: 393-399
- Pretreatment nomogram that predicts 5-year probability of metastasis following three-dimensional conformal radiation therapy for localized prostate cancer.J Clin Oncol. 2003; 21: 4568-4571
- Time to detectable metastatic disease in patients with rising prostate-specific antigen values following surgery or radiation therapy.Clin Cancer Res. 2005; 11: 8669-8673
- Pattern of prostate-specific antigen (PSA) failure dictates the probability of a positive bone scan in patients with an increasing PSA after radical prostatectomy.J Clin Oncol. 2005; 23: 1962-1968
- Pre-treatment nomogram for disease-specific survival of patients with chemotherapy-naive androgen independent prostate cancer.Eur Urol. 2006; 49: 666-674
- Nomogram for overall survival of patients with progressive metastatic prostate cancer after castration.J Clin Oncol. 2002; 20: 3972-3982
- Prostate cancer-specific survival in men treated with hormonal therapy after failure of radical prostatectomy.Eur Urol. 2007; 52: 446-454
- Prognostic model for predicting survival in men with hormone-refractory metastatic prostate cancer.J Clin Oncol. 2003; 21: 1232-1237
Article info
Publication history
Published online: July 27, 2009
Identification
Copyright
© 2009 European Association of Urology. Published by Elsevier Inc. All rights reserved.