Today, we are going to be talking about
Seamless Phase II/III Trials.
My name is Elizabeth Garrett-Mayer
and I'm an oncology biostatistician
at the Hollings Cancer Center
and the Department of Public Health Sciences
at the Medical University of South Carolina.
Along with me is Nathaniel O'Connell,
who is a graduate student
in the Department of Public Health Sciences.
I will be giving the beginning part
of this talk,
and then towards the end, we will switch over
and Nathaniel will finish the talk
by going through some case study examples.
There are traditionally
three phases of clinical trials.
The first phase is often called
a dose-finding study,
where the goal is to find the optimal dose.
These are commonly "first in man" trials
and there are a range of doses
that are evaluated to find the highest dose
that falls below,
or within, an acceptability threshold.
These trials usually have relatively
and are completed as the first part
of the clinical development of a drug.
Moving on from phase I, after an optimal dose
is found we go into the phase II setting.
At this stage, the safety of the drug
is further investigated
and the efficacy of the drug
is also investigated.
These are more exploratory in nature,
where we're looking to find
what the correct dose,
the regimen, or the schedule of the drug is.
And they may investigate early efficacy
across several dose levels,
or treatments, or even regimens,
In this phase is where we're looking
for the hint of efficacy
to see whether or not
the drug is showing sufficient promise
to be moved forward to the next phase,
which is phase III.
In phase II we often allow
relatively small sample sizes
recognizing that we need preliminary evidence
and it's not considered
a definitive phase of drug development.
The last phase of drug development
in most contexts is phase III.
In phase III it is a confirmatory stage
where efficacy is determined, often
versus a standard of care regimen.
In these trials there is a large sample size
relative to the phase II,
and they are very often multi-site studies.
Almost always are we going to be comparing
a treatment to some kind of control group;
And in this case we almost always have
relatively strict error rates,
meaning we need to have
high statistical power
and a low alpha level
also known as the type 1 error rate.
And this is because this stage is
when there is the opportunity to have
the drug approved by a federal agency,
such as in the United States,
the Federal Drug Administration.