Predicting Prostate Cancer Outcomes With Genomic Testing

Gerald Chodak, MD


November 22, 2017

Hello. I'm Dr Gerald Chodak from Medscape. Today's topic is the Genomic Prostate Score (GPS) and its potential to help men decide what to do if they have been diagnosed with a clinically localized prostate cancer. The GPS is a 17-gene test for 12 cancer genes and five reference genes. The resulting score ranges from 0 to 100; the higher the number, the greater the likelihood of the event occurring.

Van den Eeden and colleagues[1] published a report of 279 men diagnosed [with prostate cancer] between 1995 and 2010 in the Kaiser Permanente Health System who had been treated with radical prostatectomy. These men were selected out of a population of about 6000 because they had all of the information available, making it possible to do this analysis. Paraffin blocks were obtained from these patients and the GPS was performed.

Results were compared with both Cancer of the Prostate Risk Assessment (CAPRA) scores and National Comprehensive Cancer Network (NCCN) risk categories. They found that if a man had very-low-risk or low-risk prostate cancer and a GPS score < 20, none of them developed metastatic disease or died of their cancer in the next 10 years. They found that relative to CAPRA for predicting death from prostate cancer, the area under the curve (AUC) was 0.78. It increased to 0.84 using the GPS score. In other words, [GPS] improved the accuracy. When comparing GPS with CAPRA for metastatic disease, GPS improved [the accuracy] from 0.65 to 0.73. The results were slightly better when comparing it with NCCN.

What does that tell us and what does it mean? Let's start with the very-low- and low-risk category. The odds of developing metastatic disease only go up to about 20%, even at the highest GPS score of nearly 100. The question is: What fraction of men actually fall into this higher-risk category to make the test worthwhile? Under current guidelines, nearly all of these men are going to be advised to consider active surveillance. Will the GPS score change that in any meaningful way? It is going to be a rare patient who will find some benefit. What about in the high-risk category? Here it is unlikely that the GPS score will change a patient from aggressive therapy to conservative therapy. One could make an argument that it does not serve a very useful purpose for high-risk patients.

Those who might benefit are intermediate-risk patients, where there is a question of whether to consider active surveillance. A low GPS score might tip the balance in favor of that direction. This test might be useful to consider in a man who is on the fence about whether aggressive therapy is right for him.

How many patients are actually being helped by the use of this test? Do the results justify using it for far more patients? The trouble remains that although the AUC is slightly better compared with the CAPRA score, it's not that much better. It still has a 16% inaccuracy for predicting death from prostate cancer, and nearly a 25% inaccuracy for predicting the development of metastatic disease.

There is no doubt that these genetic tests are gaining in their ability to help with what we can do and what we can predict. At the end of the day, it's an estimate. It's a probability. It's not an exact number. I'm not sure that it's worth getting the information if you tell a patient they have a 25% or 30% risk of developing metastatic disease when using the GPS test and predicting maybe 10% or 5% lower if using one of the other methods, such as CAPRA or NCCN. There is the expense of the test, and at the end of the day, are enough patients going to really benefit from this method to justify its widespread use? I'm not sure that will happen.

Last, we need prospective data. We need to look at patients who used the test to make decisions and see how often the outcome was as was predicted. Until that occurs, we are going to be left with some uncertainty about what these tests really do in terms of helping patients.

I look forward to your comments. Thank you.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.