Edited from PCRI Insights November 2005 vol. 8 no. 4
By Glenn Tisman, MD
Many times during the course of my oncology practice I have been made to feel like a referee of a dispute between the urological surgeon and the radiation therapist as they vie for center stage in the treatment of the prostate cancer patient. I often see these patients for a second and sometimes a third opinion for therapeutic recommendations. These patients are generally confused about what is the best therapy since many times they have been given advice based on expert physician recommendations and they are amazed that these recommendations rendered by reasonable and honest physicians may be in conflict and even diametrically opposed.
I don’t pretend to be a know it all, and my memory for statistical outcomes is far from perfect. However, I found a solution that helps me offer patients objective treatment advice based on academic clinical studies. The techniques I use employ nomograms, the construction of which is based on statistical analysis of the outcomes of clinical trials. The clinical trials frequently involve hundreds of patients, thus lending the credibility of numbers to the results. The accuracy of nomograms has been shown to far exceed the expertise of physician professional opinion, which is often based on physician bias and less than objective parameters. I do not mean to imply that physician expertise is not of value. Physicians bring to the table medical and scientific expertise plus practical advice that can help integrate medical, social and economic realities that patients must consider while enduring various therapies. But nomograms can be a valuable tool.
Most nomograms used in clinical medicine are linear graphical pictures that require the user to draw perpendicular lines connecting other scaled parallel lines. The point of intersection yields a numeric result. This inherently carries out a specific mathematical calculation (similar to the old slide rule calculation). The intersection point represents the solution of a statistical function based on a prediction model derived from the outcome of a clinical study. The statistical equation solved by connecting the lines is frequently a regression equation i.e. Cox multivariate model or logistic regression, which is a mathematical approach to prediction. By using a nomogram, the user no longer needs a calculator or computer to solve a complex equation.
I prefer the nomogram as a prediction tool to so-called grouped analysis that attempts to fit a patient into strictly defined groups of patients with similar but not exactly the same disease characteristics. The groups frequently have overlapping clinical parameters. Examples of grouped analysis include grouping patients with high PSA values, high Gleason scores and high clinical stages as patients with high-risk prostate cancer. The problem with such groupings is that some patients placed into a poor prognostic group may have only one or two of three poor prognostic features. Such patients are bound to do better than those with all three poor prognostic features. Grouped analysis does not allow for the continuous valuation of a variable. Methods of outcome predictions using prognostic groupings are always less accurate than nomogram analysis because the use of nomograms allows for the continuous quantification of each separate disease-related variable. For example, PSA is weighted to be any number from 1 through infinity in a nomogram rather than PSA 1-10 = Weight-1 or PSA 11-20 = Weight-2 or PSA 21-30 = Weight-3 … etc. as in a grouped analysis.
A search of the medical literature reveals many useful nomograms for making prostate cancer outcome predictions. At this point we can gain a clearer understanding of how nomograms may be applied by presenting different clinical scenarios for analysis. I hope that they will be of value to patients and possibly their physicians as they both strive for the best individualized treatment plan.
The article describes eight different scenarios of situations often faced by prostate cancer patients:
Clearly, using nomograms makes readily available objective estimates of patient outcomes supported by experience gained from hundreds of previously treated patients entered into clinical trials. Their use inherently sidelines subjective physician bias, but should not void physician expertise altogether. Nomogram use and physician experience should complement each other in determining the final course of action.
The nature of the nomogram is simple and does not require the user to have a calculator or computer or spend valuable time solving complex equations. Any patient interested in participating in decision-making only needs to gather the necessary input data from their medical record to apply a nomogram to his specific case.
Editor’s Note: Nomograms give probabilities, based on scientific studies done with hundreds or thousands of patients. While these may provide some guidance, remember that they may not be based on treatments using techniques as practiced today (such as IMRT). The PCRI Helpline staff can calculate the algorithms for you but must disclaim responsibility for the accuracy or any subsequent use of the results. A patient should discuss these with his own physicians.