Wednesday, November 25, 2015

Newspapers Should Not Give Medical Advice

In an editorial, yes an editorial, in the NY Times they state:

An editorial in the journal (JAMA)  suggested that the pendulum has swung too far against prostate cancer screening and that it is time to focus on ways to use the screening test more effectively, perhaps by reserving it for men at high risk based on such factors as race, ethnicity, age, family history and the results of rectal exams. Men facing only an average risk of dying from prostate cancer might simply have blood tests and possibly biopsies every other year and undergo surgery or radiation only if there is evidence that they have a tumor that is growing and becoming more aggressive.

The problem is identifying risk.  We try Bayesian risk analyses. Namely if a patient has a family member with an aggressive form of PCa and that is a first degree relative then we assign that patient as high risk. Sometimes that works, but not always. The real risk is the genetic profile of the cancer cells distant from the prostate. But how do we find them, and in fact what specific profiles are we looking for? Frankly there are no answers.

Recently we examined a patient profile that was confusing to any Bayesian, and I am one. Let me reiterate it simply:

1. The Pt had a father with a highly aggressive form leading to death. From a Bayesian perspective this gave the Pt a 20-30% chance.

2. The Pt had a biopsy with HGPIN. This raised the risk to 50%.

3.  The HGPIN was no longer apparent on a second 24 core US biopsy. This lowered the risk to 20-30% again.

4. The Pt saw a PSA velocity well above the maximum, now raising PCa risk to 50%+.

5. The Pt was assayed with a 4K test which resulted in a less than 1 percent range.

6. However a concomitant MRI in anticipation of a biopsy revealed 3 small lesions. This raised the Bayesian risk to 75 percent.

7. An integrated MRI/US 24 core biopsy was all normal and this reduced the Bayesian risk back to  a less than 1 percent range.

This Bayesian approach shows the problems, not only with PSA, but what we understand about a priori data and its influence. The suggestion above from the Times:

Men facing only an average risk of dying from prostate cancer might simply have blood tests and possibly biopsies every other year and undergo surgery or radiation only if there is evidence that they have a tumor that is growing and becoming more aggressive.

is in my opinion based upon years of extensive review and analysis a nonsense proposal. Why? Simply, the literature is full of the complexity of PCa genomics. PCa can metastasize in less than three months in certain cases. It can enter the blood stream and seek a nice home in the bone. Then the game is lost. Why is an editorial board even suggesting this without any detailed evidence. In fact the evidence is still conflicted. So it is best to keep quiet or follow a conservative path, even for a liberal paper.