Page 140 - An Evidence Review of Active Surveillance in Men With Localized Prostate Cancer

information is available. Thus, patients having surgery are staged more accurately than those
with clinical or imaging staging alone. This bias could be reduced if the SEER database
maintained the staging information that is available prior to surgery, so that researchers can
analyze unbiased data about staging.

Key Question 2. Definition of Active Surveillance

Little new research per se is needed to address how AS has been defined by researchers.
However, interpretation of future studies would be best served if there were a standard, agreed-
upon definition of AS that clearly distinguishes it from WW and other forms of withheld or
noncurative treatments. A consensus conference may be the most appropriate forum to define
AS. Features of the definition will need to include 1) the goal or intent of the intervention (e.g.,
delaying curative treatment until there is evidence of progression); 2) the “eligibility criteria,” a
determination of which patients should be offered AS based on disease and patient
characteristics; 3) the “followup protocol,” the minimum set of tests that should be followed
(e.g., DRE and PSA), and their timing; and 4) criteria or triggers for stopping AS, when there has
been sufficient or rapid enough progression to warrant active, curative treatment.
Working under the (still unproven) assumption that AS is a safe and effective treatment
alternative to RP or RT, the best AS protocol should be investigated by randomized or other
prospective comparative studies that directly compare different protocols. The current
retrospective or case series studies provide some data to allow for comparison of protocols, but
these data are largely incomplete and adjustment using techniques of multivariable analysis is
likely inadequate to control for confounding and other biases. Examples of comparisons for
future trials could include use of different combinations of followup testing (e.g., PSA, DRE,
imaging, rebiopsy), different timing for the tests (e.g., every 3 or 6 months), and different
definitions of progression that would determine when curative treatment is offered. These trials
will require long-term followup. The outcomes of greatest clinical importance are those that are
most pertinent to patients’ health, well-being, and longevity. Examples include all-cause
mortality, prostate-cancer-specific mortality, development of symptomatic disease or distant
metastases, urological and other complications (from testing, treatment, or deferring treatment),
quality of life, anxiety, and family dynamics. Also of interest would be overall costs, use of
resources, and numbers of negative invasive tests (i.e., biopsies showing no progression that
arguably were thus unnecessary). Since only about half of men on AS require treatment due to
disease progression within 5 years, 168 and only a percentage of them will have clinically
important outcomes (e.g., cancer death), a trial may also need to be quite large to be adequately
powered.
Another related question of interest that was not asked for in the Key Question (and thus was
not systematically reviewed) is which tests are the best predictors of either progression or clinical
outcomes. Ideally, for the purpose of developing an AS protocol, these studies should be
conducted only in men who are being followed with AS, excluding men with more advanced
disease at baseline or who are undergoing curative treatment. Studies would need to properly
account for whether ADT is being used. Prospective studies that directly compare specific tests
(e.g., PSA, DRE, imaging, rebiopsy) would be most reliable. However, such studies may have
large amounts of confounding and colinearity. For example, there will likely be large variation in
the frequency of specific tests, which may be confounded with the tests themselves; the results of
some tests (e.g., DRE) may affect the frequency or use of other tests (e.g., rebiopsy), and it may

86