54 - Adaptive Designs in Early Trials (S4E9)

All right, so we all know just how important
it is to get those new medicines out there, especially
for the people who really need them, and fast,
right? But sometimes it feels like the traditional
ways of testing new drugs, especially in those
initial phases, it's kind of like we're stuck.
You know, we had to stick to a rigid plan, even
when the data is like waving its hands saying,
hey, there's a better way to do this. So for
you, the listeners, the ones who always want
to learn things efficiently, cut to the chase.
Well, the deep dive, this is going to be right
up your alley. We're tackling a real game changer.
You got it. Today, we're going. deep, really
deep into adaptive designs in those early clinical
trials. And we've got a whole stack of material
here about, you know, drug development, all the
regulations. It can get pretty dense. So our
mission today to pull out the real gems, the
core knowledge about how these innovative trial
designs can make those early testing phases way
more efficient. Because let's face it, no one
wants to waste time when it comes to potentially
life saving treatments. You're speaking my language.
Efficiency is key. So break it down for us. Adaptive
trial designs. What's the big idea? Okay, imagine
this. Instead of like setting everything in stone
at the very beginning of a trial, you build in
some flexibility points, right? So you can make
changes based on the data that's coming in as
the study goes along. Basically, you're creating
a trial that learns as it goes. Pretty cool,
huh? It's like a trial that evolves. That's got
to be way better than just sticking to the old
rigid plan. So the route this Deep dive. I'm
guessing we'll be getting into the nitty gritty
of how these trials adapt, right? Examples from
the real world, all that good stuff. Exactly.
We'll dig into the specifics, pull in some relevant
examples from those sources you shared, and of
course, we'll touch on the all -important regulatory
aspects. You know, the FDA, the ICH, got to make
sure everything's above board, our aim, to give
you a clear picture of this whole adaptive design
thing without getting bogged down in the weeds.
Sounds good to me. Now, to kick things off, Let's
talk about how these adaptive designs, how they
actually compare to those traditional fixed designs.
You know, the ones that are kind of like, well,
fixed, no room for change. Great question. So
picture a traditional fixed design. Everything
is set from the get -go. How many people you're
going to include in the trial, the doses of the
drug, who gets to participate, all of it. And
that's how it stays throughout the entire trial.
No wiggle room. And adaptive design, though,
it's like you're building in these checkpoints
where you can look at the data and say, OK, based
on what we're seeing, maybe we should tweak this
or that. It's about being smart and making adjustments
along the way. So it's almost like building in
a little bit of that choose your own adventure
aspect to the trial, but based on real hard data,
not just like flipping a coin. You got it. And
there are a bunch of ways these trials can be
designed to adapt. Take, for example, how we
figure out the right dose of a drug. I'm sure
you've heard of the classic three plus three
design, often used for those early cancer drug
trials. Yeah, vaguely. Well, our source material
on phase one and two trials, it goes into this
three plus three design. It basically involves
testing a low dose in three patients. If none
of them have any serious bad reactions, you bump
up the dose for the next three. If one patient
has a really bad reaction, then you bring in
three more at the same dose. And if two or more
have severe toxicity, well, that dose is a no
-go. It's considered too high. Right. So very
step -by -step. Not much room for, well, adapting.
So how does an adaptive approach change things
up? How does it improve on finding the right
dose? Well, think of it this way. An adaptive
design for dose finding it can be way more dynamic.
Instead of those fixed groups of three, it adjusts
the number of patients at each dose level based
on how safe the drug seems to be based on the
data. And what's really cool, it can use these
statistical models, right, like to predict the
chances of toxicity at different doses, taking
into account the data from all the patients so
far. This way we might be able to find that maximum
tolerated dose, you know, the highest dose that's
still safe, much quicker, and importantly, with
fewer patients overall. It's like using a GPS
to navigate that dose finding journey. That's
really fascinating. So you're using all this
incoming data to basically steer the ship to
make those dose adjustments as you go along.
Now I'm curious, what else can we adapt in a
trial? Well, another critical piece of the puzzle
is something called sample size adjustment. You
see, in a traditional trial, You figure out how
many participants you need right at the beginning.
And this is based on wanting to have enough statistical
power. Basically, the ability to confidently
say, hey, this drug really works if it truly
does. But here's the thing. As the trial goes
on and you start to see some early results, you
might realize that the drug is either working
much better than you thought or maybe not as
well as you hoped. And I'm guessing that can
really impact how many people you actually need
in the study to get a clear answer. Yeah. Exactly,
and that's where Adaptive Design shine. They
can build in these pre -planned checkpoints,
you know, interim analyses where you take a look
at the data. If it looks like the treatment is
a home run, really effective, you might be able
to reduce the number of participants you initially
planned for. And you can still have enough data
to confidently say whether or not it works. On
the other hand, if the drug's effect seems a
little, well, underwhelming, you might have the
option to... Increase the sample size, bring
in more people to make sure you can actually
detect a real difference if one exists. This
way you avoid having way too many people in the
trial when you don't need them or not having
enough to get a clear picture. Efficiency all
around. It makes total sense you're being smart
with your resources and importantly with the
patients who are taking part in these trials.
Now what about those different treatment options?
Can adaptive designs help us narrow down which
ones are worth pursuing? Absolutely. In some
of those early phase trials, you might be testing
a few different treatments at once, or maybe
slightly different versions of the same treatment.
And guess what? An adaptive design can be set
up to drop any treatments that are clearly not
working. You know, at those pre -planned checkpoints
where you're analyzing the data, this way, All
the effort, the money, the patients, everything
gets focused on the treatments that show real
promise. Fail fast, move on, and put those resources
where they count. Like a little competition within
the trial. And the data helps you pick the winners
early on. I like it. That's a great way to think
about it. And there's another really cool adaptation.
It's called patient population enrichment. Sometimes,
you might find that a treatment works really
well for a specific group of patients, but not
so much for others. Oh, I see. Like maybe a certain
gene makes some people respond better to a drug.
That's a perfect example. With an adaptive design,
you can collect data on those biomarkers, those
genetic markers, from the first patients in the
trial. And if that interim analysis shows that
people with a certain biomarker are the ones
who really benefit, well, you can adjust the
trial to enroll more patients from that specific
group. This makes it more likely that you'll
be able to prove the drug works in the people
who actually benefit from it. And it sets the
stage for more personalized, targeted treatments
in the future. that's the direction we want to
be heading in, right? Absolutely. Personalized
medicine is a huge deal. Now, are there any other
adaptations we should be aware of? Just one more,
and it's a little less common. It's called endpoint
modification. Basically, in some cases, the early
data might suggest that the way you're measuring
the drug's effect, that endpoint, isn't the best
one. Maybe there's a different way to measure
it that would give you a clearer picture. Now,
changing the endpoint mid -trial, It's not a
decision to be taken lightly. You got to be super
careful with the statistics to make sure the
trial is still valid. But it is a potential adaptation
you might see in some adaptive designs. OK, that
makes sense. So we've covered a lot of ground
here. We've talked about how adaptive designs
can adjust the dose, the sample size, the treatments
being tested, the patient population, even the
endpoint. Now, I remember you mentioned using
our source materials, but while those sources
might not give us specific examples of recent
adaptive trials, can we still see those core
principles of adaptive design kind of woven in
to the broader ideas in those materials? Oh,
for sure. It's all connected. For instance, you'll
notice a big theme across several of those sources.
They all talk about how crucial it is to make
drug development more efficient, you know, less
time, less money, especially those books from
season two and season three. about developing
cancer drugs. They really highlight how costly
and time consuming it can be to get new cancer
treatments to the people who desperately need
them. Adaptive designs, they directly address
this issue. They give us a way to streamline
that early research phase, make it more informative,
faster. If we can get solid data quicker and
with fewer resources, well, that's a win for
everyone. I'm with you there. Any way to speed
up the process of getting new treatments to people
while still making sure they're safe and effective,
that's a good thing. Exactly. And if you look
at that anti -cancer drug development guide we
have, they specifically call out PKPD -guided
dose escalation as being a very attractive approach.
Now, this might not be a full -blown adaptive
design with all those formal decision rules,
but it's still using the same fundamental idea.
It's about using the data you're gathering about
how the drug moves through the body and what
it does to make smart adjustments during the
trial. It's all about letting the data guide
your decisions. I see. So even before we had
this formal framework for adaptive designs, researchers
were already recognizing that, hey, using real
-time data to inform those next steps, that's
a smart way to go. Exactly. And then if we look
at the clinical trials handbook, they talk about
interim analyses, but mainly in the context of
those later stage phase three trials. But again,
it shows that fundamental idea of using the data
you're collecting to make big decisions about
the study. Like, should we stop the trial early
because the drug's a clear winner or a clear
loser? Should we make other changes to the protocol?
Adaptive designs, they essentially bring this
same approach, this idea of using interim data
to make decisions into those earlier foetas of
research, the exploratory ones. But the difference
is they build in that flexibility from the very
beginning. It's like planning for the unexpected
in a smart way. Makes sense. Now, a huge part
of any clinical trial is all the regulations,
right? So how do those big regulatory bodies
like the FDA and ICH, how did they feel about
these adaptive designs? Anything specific we
should know about? Well, it's super important
to understand that even though these adaptive
designs offer a lot of flexibility and efficiency,
they still got to play by the same rules as any
other clinical trial. The FDA, they're the ones
who review and approve all trial designs, including
these adaptive ones. And they're generally open
to innovative approaches, you know, things that
could speed up drug development. But, and this
is a big but, those designs have to be scientifically
sound. They can't cut corners. They got to make
sure the results of the trial are still reliable.
It's not like you can just say, oh, we're feeling
spontaneous, let's change things up mid -trial.
No way. There are rules. There are expectations.
One of the key principles of these adaptive designs
is something called pre -specification. And that
basically means that any potential changes, you
know, what can be changed, what data you're going
to use to decide whether to make a change, the
statistical methods you're going to use, all
of that, it has to be spelled out in the trial
protocol before the trial even starts. No surprises.
This is super important to make sure there's
no bias creeping in and that the results of the
trial are trustworthy, you know, like that you
can actually believe what the data is telling
you. And if you look at our sources on trial
design, like the practical guide and fundamentals,
they're always stressing how crucial it is to
have a detailed protocol and to stick to it.
And with adaptive designs, well, that becomes
even more important. It's like having a roadmap.
but a roadmap that includes different routes
depending on what you find along the way. But
you still need that map from the start. Okay,
what about those ICH guidelines? Where do they
fit in? Well, the ICH guidelines, as we saw in
those Season 5 materials, they're all about making
sure the regulatory standards are the same. cross
the board internationally. It's about making
sure those medicines are safe, they work, and
they're high quality no matter where you are
in the world. Now, you might not find a specific
section in those guidelines that's all about
adaptive designs, but the core ideas they're
focused on, you know, things like... good data,
patient safety, ethical research, all of that
applies to adaptive trials as well. So anyone
doing these adaptive trials, they've got to follow
all those relevant ICH guidelines for conducting
the trial, monitoring it, reporting the results,
all of it. Those guidelines, they're there to
make sure that even with all the flexibility
that adaptive designs offer, the trial is still
done in a way that's scientifically and ethically
sound. No compromises there. I see. So the regulators,
they're definitely on board with the potential
of these adaptive designs. But they want to make
sure that everything is planned out carefully
the statistics are solid and the ethics are Impeccable
right from the get -go. You got it. It's all
about that pre -specification It keeps everyone
honest, you know, it shows that the researchers
aren't just making things up as they go along
Makes sense. So big picture What are those main
advantages of adaptive designs that we've talked
about just to sum it all up? Sure, the biggest
wins with adaptive designs are all about being
more efficient They give you a chance to find
those effective doses quicker, to use your resources
wisely by adjusting the number of people in the
trial, to drop those treatments that are clearly
duds early on, and to really zero in on the patients
who are most likely to benefit. And at the end
of the day, that efficiency means we can potentially
get those life -saving treatments out there faster.
And that ties back to what we talked about at
the very beginning, right? About wanting to learn
things quickly and efficiently. Well... Adaptive
designs, they can help us do just that. And that
efficiency probably helps with the cost, too,
right? Bringing those costs down. Absolutely.
Saving time, saving money, those are huge benefits.
But it's important to remember that adaptive
designs, they're not a walk in the park. They
need a lot of careful planning, some pretty sophisticated
statistical analysis, and they got to be executed
flawlessly. They sound like incredibly powerful
tools, but they definitely require some serious
expertise to do them right. So, as we wrap up,
any final thoughts you want to leave our listeners
with? Yeah, I want you to think about this. We're
living in a time where we have more and more
real -time data at our fingertips. And those
tools for analyzing that data, they're getting
more sophisticated all the time. So imagine how
this could change early phase trials even more.
Maybe we'll see designs that adapt even faster,
maybe even using AI, machine learning, to optimize
trials in real time. It's a really exciting thought,
and it builds on those conversations we've had
about the role of AI in drug development. The
possibilities are huge. That's mind -blowing.
AI guiding those clinical trials, making them
even smarter. The future of drug development
is looking pretty amazing. Well, I want to thank
you for joining me on this deep dive into adaptive
designs. It's been a fascinating conversation.
The pleasure was all mine. Keep exploring this
whole world of drug development, all the amazing
innovations that are happening. There's always
something new to learn.