Preventive Medicine Column

Dr. David L. Katz

According to the United States Preventive Services Task Force, PSA- which actually stands for ‘prostate specific antigen’- might as well mean ‘please stop asking!’  Please stop asking whether or not men have prostate cancer by using this test, in other words, because it is tantamount to asking for trouble.

The news that that highly regarded Task Force now officially recommends against screening for prostate cancer has already proved upsetting to some, and will doubtless prove so to many others as they learn of the new report. There are people I know, perhaps people you know, and some celebrities we all know who feel their lives, or the lives of loved ones, have been saved by prostate cancer screening with the PSA test.

Can they, and the Task Force, both be right?  Paradoxical as it may seem, the answer is: yes!

For starters, the prostate specific antigen is a protein specific to the prostate, but not to prostate cancer.  Levels of PSA in the blood vary in normal, healthy men; and can rise with benign enlargement of the prostate.  These values overlap the values seen with prostate cancer.

A truly specific test is one that only generates an abnormal result when the disease in question is present.  Such a test is very useful at ruling disease in, because it will almost never be positive when disease is absent- and thus if positive, the test reliably rules in disease.  Alas, the PSA is not such a test.  The result of the test’s low specificity is that many men without prostate cancer wind up having unnecessary biopsies, and sometimes, complications of those biopsies.

The PSA might still be useful if it were highly sensitive.  A highly sensitive test is reliably abnormal when disease is present.  When a highly sensitive test result is normal, it is a very reliable indicator that disease is truly absent.  But again, the PSA falls short.  Levels in many cases of prostate cancer are not appreciably different from levels seen in normal men- and thus the test cannot be counted on to rule disease out either.

But the most important failing of the PSA is not so much about its use in detecting disease, but rather the problem of knowing what to do about prostate cancer when it is found.

Cancer screening, or for that matter any disease screening, is, in the most literal sense, looking for trouble.  Looking for trouble makes sense if by finding it early you can fix it.   But if you don’t know what to do with the trouble you find, you are no longer just looking for trouble- you are asking for trouble, too.

Prostate cancer is unpredictable.  More often than not, it does not grow or progress.  In fact, as many as 80% of men who live past age 80 have prostate cancer when they die and most never know it.  But some prostate cancers grow, spread, and kill.  Thousands of men die of prostate cancer each year in the U.S.

The real trouble with looking for this particular trouble is that we can’t reliably tell these varieties apart.  The result is that PSA screening results in treatment for prostate cancer that doesn’t need treatment.  And the result of that is avoidable harms, including impotence, dysfunction of bladder and bowel, post-operative infections, and so on.

I am a preventive medicine specialist.  I also happen to be a man well past 40 who owns a prostate- and to make this personal, I have NOT opted for prostate cancer screening, because I am familiar with the evidence on which the USPSTF is basing its current recommendation.  I am, of course, also familiar with the high-profile stories of PSA testing seemingly saving lives.  How to reconcile those two?

Probabilistically.  PSA screening may, in fact, save lives.  But the question is: is it more likely to save my life, or cause me harm?

There are specific calculations to help sort out just such dilemmas.  The most useful and intuitive are the number needed to treat (NNT), and number needed to harm (NNH).  The former measure indicates how many people need to undergo a treatment or test, and face its potential risks, before one person actually benefits.  The latter says how many people are harmed for every one who is helped.

There are various published estimates of the NNT and NNH for prostate cancer screening.  The Task Force conclusion is simply that the NNT is too high, and the NNH too low to justify routine use of the test.

Which brings us to the most uncomfortable part of all this.  Isn’t the Task Force trading off the lives of individuals who could be saved against a less important ‘population level’ effect?

It may feel that way, but it’s not so.  Let’s illuminate why through the time-honored technique of reduction ad absurdum.

Occasionally, someone with no prior signs of heart disease experiences sudden cardiac death that coronary bypass surgery might have prevented.  We might, therefore, simply open up the chests of everyone at age 25, or 30, or 35, on the chance that they were the rare individual facing such a fate.  The gummed up coronaries could be bypassed, probably saving a life- while everyone else’s chest would simply be closed up again.  Alternatively, we know some people develop advanced cancer without showing signs of it in early stages.  We might simply give everyone chemotherapy to make sure such cases don’t sneak by.

I am confident no one would sanction such approaches.  Even knowing I might be the rare individual prone to sudden cardiac death, I would not want to undergo open heart surgery ‘just in case,’ and I would certainly not want to impose that approach on anyone else.

PSA screening as a matter of routine is, of course, far less extreme- but the difference is one of degree, not kind.  For a screening test to be useful, it must be applied widely, so the relatively uncommon person who actually has the disease is found.  We don’t know in advance who that person is- so we have to go looking for him in a crowd.  But the result of such widespread application is that many people will be exposed to the test- and whatever harms are attached to it.

For better or worse, then, how we care well for any one body is inextricably tied up with implications for the body politic. It simply doesn’t make sense to impose on a population an approach that does more harm than good.

This is not a statement of social or political preference, and has nothing to do with ideology.  It is a matter of statistical probability.  When the result of testing is net harm, any one person tested is more apt to be harmed than helped.  The Task Force relies on data, and boldly confronts the implications.  Their conclusions may inspire passion in others, but the conclusions themselves are dispassionate and objective.

Tempting as it may be to oppose an evidence-based conclusion with passion, conviction, and wishful thinking- we do so at our collective, and individual peril.




Dr. David L. Katz;