We regularly see statements that almost all men will show evidence of
prostate cancer if you look hard enough.

The basis of this belief is that autopsy studies (and there haven't been
many) have shown that many men show evidence of prostate cancer when
their prostates are examined after death, and the percentages go up with
age.  There are some potential selection bias problems with such studies
because autopsies are not routinely done on all men, and the results of
different studies differ significantly.  But the more relevant question
is how many of these cancers found on autopsy would have been clinically
significant if the men had lived longer.

In living men, a biopsy is ordered if PSA rises above a certain level,
or rises too fast, or if a digitial rectal examination reveals something
unusual.   If the biopsy shows evidence of prostate cancer, and the
Gleason score is 5 or more (which is true in the overhwhelming majority
of such cases), then that is considered a clinically significant case of
prostate cancer and will usually lead to treatment, at least in men
under 70.  The question is how many of those cancers are really
innocuous and need not be treated.

Now it turns out that there was a recent study which gave us some
information about that.  The PCPT study was interested in whether
finasteride could help avoid prostate cancer.  So they administered that
drug to a bunch of men with "normal" PSA using the current criterion at
the time and also to a bunch of controls.  They found that it did appear
to have a protective effect but it also appeared to result in more
aggressive cancers in men using the drug, a somewhat mixed result.

But they also had the control group who were given a placebo drug but
also were biopsied.   They found that 15 percent of those men showed
evidence of prostate cancer, but only 2.3 percent of them had tumors of
Gleason grade 8 or more.   Most of them were Gleason 5 or 6.

Now there are some caveats about this.  Certainly some of these moderate
range Gleason grade cancers they found were significant and would
ultimately develop into metastatic prostate cancer if untreated,
despite the relatively low PSA levels.  And for men with higher PSA
values, it seems plausible that the number of clinically insignificant
prosate cancers would be lower.   But it does seem that we can now take
15 percent as a ball park upper bound for the percentage of cancers now
being treated by radiation and surgery which don't actually need to be
treated, at least in men with a long enough lifespan for an early
prostate cancer to develop into advanced metastatic cancer.

Interestingly enough, an earlier study based on Medicare data came up
with a similar estimate, at least in men of European descent, for the
number of cases being treated unnecessarily.

I think this suggests that under current practice too many men are being
treated unnecessarily, particularly those with Gleason 5 and 6 cancers.
 But the contention that almost all men being treated are in this
category seems unlikely.  And the statement that if you did biopsies on
all men, you would almost always find prostate cancer appears to be
false.  Be that as it may, we certainly need better ways to distinguish
innocuous cancers from those needing aggressive treatment.  It is to be
hoped that methods currently under research will lead to such methods in
the next several years.  Then, perhaps this particular debate will
become moot.