It seems to me that if anything, it might suggest that with some class
of patients, low testosterone, which may just mean below normal
testosterone, is associated with more aggreesive disease but not with
PSA recurrence.  Unfortunately, as the authors readily admit, in
retrospective studies like this, it is very difficult to control for
confounding factors.  In particular, it is quite possible that to a
certain extent the more aggresive disease is causing the low
testosterone.  I haven't followed studied carefully Leibowitz's claim,
but it is my impression that his patient population os very different
from that for this study.  I think the comparison would be of "apples
and oranges".

Ron,

This result has been talked about before.  I referred to it in my
paper, citing the study published in Cancer Detect Prev 2007,
31:149-153 which showed that lower levels of T were associated with
worsening clinical staging, worsening histological staging, and more
poorly differentiated adenocarcinomas for PC.  My model predicts all
of this, since lower levels of T result in higher levels of bcl-2.
Since the cell division rate of PC seems to be about the same no
matter what the Gleason score or what the growth conditions, the
aggressiveness of the PC is inversely correlated to the rate of
apoptosis.  Since more bcl-2 means a lower rate of apoptosis, the
result is a higher Gleason score.

Now biochemical failure following surgery depends on whether or not
the gene has been activated that causes metastases.  This is
essentially a random event and independent of the bcl-2 level, and
therefore independent of the level of T.

Ed Friedman