There was a flurry of news articles over the last few weeks about the Oncotype DX prostate test (and others) which is worth thinking about. These genomic tests which are well reviewed in a recent NYT article, are based (in general) on the observations that certain genes, or sets of genes can be turned on or off more frequently in some cancers than others. By evaluating large numbers of patients, for example, with Gleason 7 cancer (intermediate risk), finding out those who live for 10 years and have recurrence or not, one can then separate the “bad” from the “good” intermediate risk cancers. The next step is to evaluate a second group of patients where the outcome is not known to the investigators and see if the gene profiles obtained on some banked tissue allowed them to correctly predict what was going to happen. Different labs/companies have come up with different panels of genes using sophisticated algorithms that allow a large gene set to be reduced to a much smaller set of genes which can then be relatively economically tested. (note I said “relatively”…$3,820 may not be so economical if you are an insurance payor or Medicare and there is no evidence that the results of the test save money…)
For all of these tests, the question is what men and their doctors will do with the results. The report that comes back to the office is often in the form of a bar graph showing the percentage of men who progress, or who die with a certain “number” that is derived from all the gene expression profiling data. It has taken several years for any of these tests to outperform the Gleason scoring system, but now many of them really do add additional information. For example one man with a Gleason 7 cancer might have a score of “23”, look on a graph and find that his likelihood of having progressive disease is “34% at 5 years, 43% at 10 years” while another man would find a score of “62” and find worse odds. If your odds are high of dying, do you then decide to do less aggressive therapy or more? If you have a Gleason 6 cancer and a low score, will it make you more likely to choose active surveillance or will the word “cancer” be enough that you want treatment regardless? And if you were going to want treatment regardless, then why did you get the test?
A more profound outcome that is already becoming reality for oncologists is the use of such testing not just for prognosticating, but for predicting responses to various therapies. Already, many of the newest “targeted drugs” are approved only in the setting where a mutation analysis or gene profile predicts that the drug will work in a given patient. In this case, the savings (both in toxicity and finances) can be huge. Why take an expensive drug for 4 months if there is a test that says it only has a 4% chance of working in you? Set against this paradigm is the remarkable heterogeneity of cancer in any one individual. Does a “negative test” really tell us about the whole cancer situation or only the situation about that small snippet of tumor removed for testing? The good news is we are making progress, and it is the exceptional research infrastructure supported by the NIH and organizations like PCF and Movember that allows it to happen.