What is going to kill me? – the cloudy crystal ball

To view this post in my blog allowing you to sign up and make comments, please click here.

With an intense focus on prostate cancer, it is easy to overlook the reality of other causes of death or disability in making decisions about therapy. An example of this issue is the proliferation of molecular tests that have been validated to separate patients with “intermediate risk”, or “low risk” into “even lower” or “even higher” risk disease categories using a number of different gene expression profiles on the tumor or biopsy material. For example, Genomic Health offers the Oncotype Dx test that provides a “Genomic Prostate Score” that gives a patient who (based on clinical criteria such as PSA and number of biopsy cores positive) falls into a low or intermediate risk category another lab value (GPS) that can potentially be useful in making a decision about treatment. GenomeDx has a test that can evaluate high risk men after prostatectomy to more accurately predict metastatic disease at 5 years. There is a very balanced article on the challenges of using these tests (which are a potential step forward to be sure) in the real world of the clinic here.

However, in all of the excitement and marketing of these and other tests, a couple of key facts are often overlooked (and may be much more important in decision making). Prostate cancer is generally a slow disease anyway. Competing mortality looms large as patients get older. And most importantly, there are validated ways to put the “whole patient” into the picture before ordering these tests, whether they be a PSA, biopsy, or molecular analysis. The Charlson comorbidity index can be extremely useful in predicting survival and is barely ever mentioned in the molecular analysis literature/reports. It is a simple yes/no answer to whether a patient has any of these 12 conditions: diabetes, bleeding gastrointestinal ulcer, chronic lung disease, congestive heart failure, stroke, myocardial infarction, angina or chest pain, cirrhosis or liver disease, arthritis, inflammatory bowel disease, hypertension, and depression. In a lovely article published last year, the use of this analysis in relationship to prostate cancer mortality gave a vivid picture of prostate cancer mortality in the larger setting of 3533 men with prostate cancer. A snapshot of their data looks like this:

Screen Shot 2014-06-19 at 9.15.54 AM

Very often, the comorbid conditions lead to death from another cause. In my opinion (and in my practice), we too often ignore our ability to quantify the risk of dying from “something else” when we focus so intensely on the PSA or other tests in counseling patients about what to do. It is also true that patient perception of test results can vary dramatically. One patient with a “GPS score” of 10 might be reassured, while another will perceive it as “not low enough” and opt for aggressive treatment rather than observation. To some extent this exposes the fallacy of “we need to separate the issue of treatment from that of diagnosis” thinking. Until the crystal ball becomes crystal clear, management of prostate cancer will remain challenging and requires the kind of wholistic thinking that is often better done by primary care physicians or public health professionals than by prostate cancer docs, or their patients.

7 Comments

Filed under General Prostate Cancer Issues

7 responses to “What is going to kill me? – the cloudy crystal ball

  1. Dennis Harrington

    To date none of the physicians I have talked with have mentioned the Oncotype test. It has been essentially which treatment is best not much about whether to treat or not. I contacted the company to get more info

  2. Dennis Harrington

    A couple of docs were familiar with its use in breast cancer but not prostate. Have you seen the clinical validation studies in prostate cancer?

  3. There are plenty of studies “validating” the molecular tests (PC3, Prolaris, Oncotype Dx, tmprss-erg, Decipher, etc. etc.) IF by “validate” one means that a test has been developed using the test on some retrospective group of patients in whom the outcome is known and the test used that data set to “learn” which genes gave prognostic or diagnostic information….and then validation occurred on another data set of patients and was also able to predict their status for the same end point. HOWEVER, in breast cancer, the use of such a test to separate patients into different TREATMENT categories with an outcome that was affected by using the predictive biomarker has been validated which is a higher bar. The discussion in the reference hypertexted at the end of paragraph one is worth reading. Doing a test and making a decision based on that test that is the right decision is what is needed. For example, making a decision to do active surveillance based on a low GPS score in one group of patients compared to just using the already available Gleason score/PSA/stage in a comparable group of patients and then finding out that in the end, the same number of patients progressed at 5 years and came off surveillance would mean that the GPS score didn’t add much.

  4. hellofromhouston

    I agree and I have some experience with genetic markers although not in oncology. I would say in general these kinds of tests are only recently and I mean very recently becoming useful in a clinical setting and also being reimbursable. Unless you as a physician can counsel a patient to use the test result to forgo a treatment option or pursue a different option based on peer reviewed scientific reproducible data it makes these tests kind of a nice to have versus really useful. In the insurance company’s eyes it is paying for a test that wont change the ultimate cost of treatment so why cover it? Unfortunately most of these companies dont have the financial strength to do through the time to accomplish the research to the highest levels.

    • Excellent perspective and right on. Plus the issue that they CAN be very cost effective if they help us NOT do something expensive that doesn’t work and home in on the patients where a drug or procedure DOES work. Then the question becomes how much we save on the non-intervention vs how much life/quality is gained from the intervention. And as we know, in the big sweep of things this is diminishing returns anyway – spending money on vaccinating kids and clean water has a huge benefit. Prolonging the life of an aging cancer patient or heart failure patient by 4 months at a cost of $100K is questionable.

  5. Marshall Freedman

    Your use of “wholistic” = whole + holistic, is exactly right!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s