Clinical Evaluation of Therapeutic Vaccines for HIV Seropositives: Approaches to the Design of Efficacy Trials

o Why is it so Important to insist on clinical endpoints? The nucleoside
analogs ddl and ddC were both approved on the basis of CD4 counts, and
for ddl we have seen this decision validated when the clinical results
became available.
Declines in CD4 counts are clearly associated with increased risk of
AIDS-defining events and death. We know, however, that the CD4 can be
modestly perturbed by many things (e.g., smoking) that have nothing to
do with affecting the true course of disease. There are no data to show
that improvements (or stabilizations) of CD4 counts induced by this
class of treatments ultimately correlate with improved clinical outcome;
therefore, we have no assurance that the types of CD4 increases seen
with AZT and ddl will have any clinical meaning. Even in this setting,
a very dramatic effect on CD4 counts (or other markers) might be
compelling; however, phase I data have not clearly demonstrated any
effects on markers, let alone dramatic ones.
It is important to consider the results of some recent trials in which
drugs that had physiological effects that seemed very likely to predict
clinical benefit were shown to actually increase mortality over that
seen with the placebo control (encainide, flecainide, moricizine,
milrinone). Given that the likeliest target population for therapeutic
vaccines is that with the longest life expectancy, such risks cannot be
disregarded.
o How will you be able to control for the effects of all the other
medications patients may be taking?
Because the study will be blinded, there is no reason to expect use of
concomitant medications to differ among arms. With a sample size as
large as planned, the probability that there will be an imbalance with
respect to use of some very effective therapy is exceedingly small. The
study would be designed to detect an effect of vaccines in a "real
world" context in which patients may in fact be taking a variety of
therapies.
o Why does the sample size have to be so large?
Sample size is driven primarily by the number of events (endpoints)
required to make a reliable comparison among treatments. Since clinical
disease progression is rarely observed in individuals with CD4 counts
over 200, large numbers of such individuals must be entered in order to
observe a sufficient number of events in a reasonable time frame.
o What if one or more of the products in the trial seems to present safety
problems? What if one can see early on that one product is far superior
to placebo, or even to the others?
The accumulating data from the trial would be regularly reviewed by an
independent data monitoring committee, whose responsibility it would be