In the Interim...

In this episode of "In the Interim…" Dr. Scott Berry is joined by Professors Victoria Cornelius, Danny McAuley, and Anthony Gordon, for a technical review of the PANTHER trial—an international, Phase 2 adaptive platform evaluating pharmacologic interventions for ARDS. The trial is open-label and does not employ blinding, as discussed in the episode. The primary endpoint is 28-day organ support-free days (death as -1, survivors 0–28 days), analyzed with a Bayesian proportional odds model. PANTHER uses  stratification by hyper- and hypoinflammatory subphenotypes, with fixed, equal randomization within each stratum. Analyses for treatments are separated by stratum, reflecting the potential of differential treatment effects. Quarterly interim analyses allow early stopping by stratum for efficacy or futility. Content includes explicit discussion of infrastructure: rapid device deployment, centralized data for trial and future biological discovery, and governance challenges in multinational collaboration. Funding is provided by NIHR (UK), US Department of Defense, CIHR (Canada), NHMRC and MRFF (Australia), HRB (Ireland), and additional support from Germany and Japan. PANTHER is positioned to streamline Phase 2 critical care drug testing and facilitate graduation to larger platforms such as REMAP-CAP, with potential to expedite pharmaceutical evaluation and accelerate ARDS therapeutic development.

Key Highlights
  • Real-time phenotyping (Randox device) to stratify ARDS patients.
  • Separate Bayesian analyses by phenotype stratum.
  • Open-label, fixed randomization within stratum.
  • 28-day organ support-free days as a composite endpoint.
  • Quarterly interim analyses enable early dropping or graduation of arms by strata.
  • Central data resource and biosample collection for future research.
  • Operational, funding, and device logistics for global trial deployment.
  • Transition of Phase 2 results to established Phase 3 platforms (e.g., REMAP-CAP).
For more, visit us at https://www.berryconsultants.com/

Creators and Guests

Host
Scott Berry
President and a Senior Statistical Scientist at Berry Consultants, LLC

What is In the Interim...?

A podcast on statistical science and clinical trials.

Explore the intricacies of Bayesian statistics and adaptive clinical trials. Uncover methods that push beyond conventional paradigms, ushering in data-driven insights that enhance trial outcomes while ensuring safety and efficacy. Join us as we dive into complex medical challenges and regulatory landscapes, offering innovative solutions tailored for pharma pioneers. Featuring expertise from industry leaders, each episode is crafted to provide clarity, foster debate, and challenge mainstream perspectives, ensuring you remain at the forefront of clinical trial excellence.

Judith: Welcome to Berry's In the
Interim podcast, where we explore the

cutting edge of innovative clinical
trial design for the pharmaceutical and

medical industries, and so much more.

Let's dive in.

Scott Berry: All right.

Welcome everybody.

Back to, in the interim, I'm your
host, Scott Berry, and today I have

three guests and we're gonna talk
about a new adaptive platform trial.

And the, the trial name is Panther and
we'll get to what, what Panther is.

But, uh, let me introduce
you to our guest today.

First, I have Professor
Victoria Cornelius.

She's a professor of medical statistics
and trial methodology, and the director

of the Imperial Imperial Clinical Unit
and co-director of the NIHR Research

Support Services, Imperial College London.

And she has extensive experience in
designing and analyzing trials to

evaluate drug and complex interventions.

And uses, uh, both Bayesian and
Frequentist adaptive designs using

innovative statistical, uh, approaches.

And we will talk about those today.

Uh, we also have Dr.

Danny McCauley, who is a consultant
and professor in intensive care

medicine at the regional intensive care
unit at the Royal Victoria Hospital

in Queens University in Belfast.

He is the NIHR Scientific
Director for research programs.

He has several research interests,
including acute respiratory distress

syndrome, which will be a topic for
today, uh, as well as recovery following

critical illness, uh, in clinical trials.

Danny and I have worked together, by
the way, extensively on the REMAP CAP

trial, a different trial that, uh, uh,
isn't the main topic, uh, of today.

And we have Professor Anthony Gordon.

Uh, uh, professor Gordon is the head
and of the division of Anesthetics

Pain Medicine and Intensive
Care at Imperial College London.

He is a consultant, uh, on the
adult intensive care unit at St.

Mary's Hospital and is an
NIHR Senior investigator.

His research focuses on developing
precision medicine and sepsis.

And he's been the chief investigator for
multiple clinical trials in critical care

and he's now the director of the UK's
NIHR Health Tech Technology Assessment

Program where he's keen to encourage
innovative and novel trial designs.

And I think we have a pretty
innovative and novel one today.

Uh, and Tony is also a member of the most
excellent order of the British Empire.

Uh, and I've also worked quite
a bit, uh, uh, for a number of

years with Tony on remap Cap.

We've all worked together, so
welcome everybody to in the interim.

Anthony Gordon: Thanks Scott.

Danny McAuley: Thanks Scott.

Scott Berry: all right, so Panther.

Uh, Panther is a new platform trial, and
I, uh, I should say that I am a member of

the DSMB and I'm honored to be a member
of the DSMB, but we have seen no data yet.

So I'm not unblinded, there's
no risk that I'm going to give

away results in the trial.

Uh, somewhat familiar with the
trial, but, uh, uh, so this,

that'll be a, a fun part of this.

So the, the Panther platform trial,
uh, and it, it labels the objective of

this is to accelerate the development
of pharmacological therapies for

critical illness by establishing an
international phase two precision

medicine adaptive platform trial,
uh, to test for efficacy, uh, in.

Uh, pharmacological intervention in
critically ill patients for acute

respiratory distress syndrome and pandemic
infection and different sub phenotypes.

So tell me about the Panther trial.

First of all, why, why do
we need the Panther trial?

What is the Panther trial
trying to address and solve?

Danny McAuley: Thanks, Scott.

I'm happy to kick off and we'll see where,
uh, we, we land with Tony and Victoria.

So I, I guess in terms of what we're
trying to do, uh, there are several

challenges, um, in treating, uh, a RDS.

Um, there's a huge amount of heterogeneity
within the, uh, overall population.

Um, and I think.

More recently, over the last sort of
five to 10 years, we've recognized

that there are phenotypes that
exist within these, uh, syndromic

definitions of, of conditions that
we deal with in intensive care.

Like IRDS the most consistently reported,
and there are others Tony may wanna

pick up on, um, are the inflammatory
phenotypes that are, uh, described

originally by, uh, Carlin Calie.

Over now, 10 years ago.

Essentially, these now have
been replicated in multiple,

uh, secondary analysis of RCTs,
observational cohorts, adult kids.

So we're trying to get at that
heterogeneity by focusing on, on

those phenotypes, and we've only
recently been able to identify.

Those at the, the bedside,
which we can come back to.

So that's one big, um, thing
that we're trying to deal with.

The second, uh, big issue is that
I think traditionally in, uh,

critical care, and particularly
ARDS we've jumped from relatively

small promising phase 2 studies.

into Big phase 3 Um, what we're
really trying to do with, uh, Panther

is give confidence in the therapies
that we think might work, uh, with a

much greater amount of, of certainty.

So move from small phase 2 to
big phase 2 within, uh, Panther.

And then I think that the
other issue is that, um.

We've really struggled to get, uh,
uh, commercial partners engaged

within, uh, the critical care
space, particularly within ARDS

and sepsis, largely because of a.

Con misconception, I think
that, um, uh, ARDS and critical

illness are are a, a graveyard.

And I think what we want to try and
do is de-risk the involvement with

commercial partners because we have an
established ongoing platform that we can

hopefully answer questions that, uh, help.

Commercial companies, uh, at a
sort of, uh, lower risk, but uh,

also more importantly and perhaps
most importantly, help patients.

So I'll, uh, stop there and
see what others have to say.

Anthony Gordon: Yeah,
Scott, I think obviously.

Align with Danny's thinking.

I think it's, we've got
critically ill patients.

There's this unmet need
to improve their outcomes.

So these are patients
at high risk of dying.

provide them generally with supportive
care, but when it's come to trials

of specific drugs that might, uh,
modulate, for instance their immune

response to that critical illness,
promising drugs from preclinical studies

haven't translated into that benefit.

And I think it is because we've.

As clinicians, we group patients
with similar signs, symptoms,

organs that aren't working.

So in this case, lung injury,

given them the treatment.

And probably some of those treatments
have had potential for benefit.

But in the trials we've just not seen it
because there are others who don't benefit

and probably some, uh, that are harmed.

And as Danny said, there's been
a big emphasis in actually.

Breaking down these clinical syndromes
into more biological sub phenotypes.

And then it would seem that the drugs,
if you could target the biology,

are more likely to be beneficial.

And so with this, um, if we
do that though, it needs to

be, uh, a joined up effort.

International is the other aspect of this.

If we try and do this on
our own, because we're.

Breaking down the populations
into smaller groups.

Um, we think the important thing is
to come together so that we can work

internationally, get answers more quickly.

Um, so that is the other key part
of this being an international,

uh, platform, getting the key
investigators to work together.

Scott Berry: Oh, okay.

So it, it, it's an area, A RDS is an
area that not much has worked in, and

yet it's still a clinical problem.

It's, it's, I don't wanna
call it a, a graveyard, but

it's been a challenging area.

And you think to some extent, part of the
challenge is ID not all a RDS is the same.

And so it may be some treatments,
precision medicine, treat some

of them well, but others that
may be negative and the net.

You run a trial, it looks like
nothing but so this trial can try to

address heterogeneity of effect, uh,
within this, do it in a larger scale.

An aspect of the platform, I
imagine is easier shots on goal.

So if you've got this up and running and
you want to bring in a treatment from a

pharmaceutical company or anything to,
to take that shot on goal, to give it

a shot to investigate it, much cheaper,
much smaller than going out and building

a whole new stadium to test that one.

I imagine that's why the platform
part of this is, is interesting.

Anthony Gordon: Yeah,
I think that's right.

And when you put it like that,
there's maybe COVID-19 gave

us some of that learning and
has helped to accelerate that.

This work was going on already.

But we saw the benefit of
platforms of bringing in drugs

rapidly and getting answers.

But I think also just underpins the
biological rationale behind this.

Uh, with COVID, a new infection,
um, was essentially a RDS, um,

but very homogeneous, particularly
at the beginning of the pandemic.

And that homogeneity, I think, was.

What enabled us, um, you know, multiple
groups of trials to show, uh, the

drugs that were successful in immune
modulation, uh, drugs, particularly

in the critically ill, uh, population.

So that sort, I think, reassures us
that, that the biological rational is

solid, that we can improve outcomes if
we identify the right patients with the,

uh, right underlying biology,
uh, with these therapies.

Scott Berry: Yep.

Alright, so before before we move forward,
a RDS, um, we've heard the term, you

know, what, what is a RDS and I know
you talked about the heterogeneity.

So I mean, what is, what is a RDS.

Danny McAuley: So, uh, essentially,
um, a RDS is severe, uh, respiratory

failure that is, uh, not due to.

Other recognized conditions like
heart failure or, uh, fluid overload.

Um, and it's characterized by, you know,
x-ray changes that are sort of bilateral

patchiness and severely impaired, um,
lung function that we, um, uh, sort of

measured using measures of oxygenation
typically, um, tho those people are on a.

Mechanical ventilator, although, uh,
more recently the, the sort of syndromic

definition has been updated to include,
um, people who are receiving high flow

nasal oxygen due to the increasing
and, and widespread use of that.

So, so that's the syndromic definition.

But I guess, you know, as you
might imagine, that sort of is

underpinned by a whole raft of
different biological processes that

we're really only now starting to.

Get at, which is the sort
of move to precision.

Scott Berry: Hmm.

Hmm.

Okay.

And, and you could this acute failure,
so this is, uh, as opposed to as

chronic, uh, uh, in a particular way.

So some, some particular event has
caused this respiratory distress,

and that could be sepsis cap and a
whole wealth of other things that

get somebody into this acute status.

Danny McAuley: Yeah, that's right.

So I mean, if you get knocked
over by a bus, you break your

leg, you have near drowning
sepsis, you have bad pancreatitis.

So, and again, if you
imagine the heterogeneity.

That as well, adding into the mix.

So it, it's not at all surprising that
we, if we've treated all these patients

the same way that we've struggled
to find a, a specific, um, therapy.

So I, I think again, that just
emphasizes the need to, to follow the

biology and the then the phenotypes.

Scott Berry: Yeah.

Okay.

So there's this larger goal we,
we, we should say, by the way,

so you're interested in this.

Panther platform trial.

And you've, you've presented
this solution to a problem and

you have gotten this funded.

Uh, so who is the major funder of this?

Danny McAuley: So, um, the, the
sort of first funder was the, uh,

national Institute for Health and
Care, uh, um, research in the uk.

Um, I should say, uh, Tony and
I also, uh, worked for the NIHR

as as well as, as you mentioned.

So, uh, just to, to mention that.

But we've now been, um.

Really successful in getting multiple
other funders, which I think highlights

the, the sort of recognition of
the, the importance of the unmet

need Were funded in the US and
were by the Department of Defense.

Were funded in, uh, Canada by the CIHR.

We're funded, uh, in Australia now
by, uh, a collaborative between

N-H-M-R-C and MRFF, and we've also
got funding in Ireland from HRB as

well as, uh, Germany and, uh, Japan.

I don't think I've left any out
there, but if I have, uh, but

Tony will correct me or Victoria.

Scott Berry: Yeah.

Okay, so we, we kind of understand
that what, what we want this to solve.

And now this sort of comes to the
clinical trial science part of it.

So how do we create something that
solves some of these problems?

So maybe we should now talk about what the
trial is, uh, the science of the trial.

So, Victoria, this is a platform trial.

Now, platform trials can mean lots
of different things, uh, in this.

Uh, what, let, let's talk
about the trial design.

What, what is the trial design now to
help create some of these solutions?

Victoria Cornelius: Uh, yeah.

Thank you Scott.

It's uh, lovely to be with
you here today as well.

Um.

So it is interesting that you said that
platforms can be lots of different things.

Um, uh, I think, yeah, it, it would be
good to talk about that and get your,

your opinion on whether they, how,
how, how different those things can be.

Ultimately, we're talking
about platform trials.

I think we're talking about, um, one
area, one clinical area, and we're

talking about the ability to adapt things
over time, to add in interventions and

to drop those interventions as well.

So that, that's how I would see it.

So some sort of adaptive.

Uh, trials design within, you
know, o overarching, uh, structure.

Um, so with this, this is exciting.

Pan, pan is exciting 'cause we've got
these, like I say, we're doing these,

uh, interest in these two phenotypes.

I dunno if we've named them
yet actually as hyper and hypo

inflammatory, uh, processes.

Um, and we, what?

The, the overarching design actually is to
stratify by, by these two sub phenotypes.

So I would call it a stratified,
uh, trial, uh, adaptive.

Um, and within that, um, we are gonna
be using, we've, we've chosen to do this

within a analysis framework as well.

Scott Berry: Hmm.

Okay.

So there's, uh, by, by the way,
one of the fun challenges of these

podcasts is not having slides.

Uh, or you could have a slide
that shows all of these components

together, but you have to sort
of describe it to the audience.

So the platform part of this is you're
gonna bring in multiple therapies,

and right now you have two of them
that you have brought into the

trial and we can talk more about,
but Simvastatin and Baricitinib.

So you have the two arms
in remap cap, for example.

We might do factorial randomization, where
you could get one or the other or both.

In this trial, I think at least now you've
got this setup where you have parallel

arms, so you have a common control arm.

You can have simvastatin and Baricitinib,
and when you add a new arm, it kind

of enters in this parallel way.

So a patient that comes
in is randomized among.

Today, right now with the, with those
two experimental arms, randomized

among those three possibilities.

Is that right?

Victoria Cornelius: That's right.

But stratified obviously
with, because we are

doing it in the hypo inflammatory
and the hyper inflammatory as well.

Yeah.

And we, yeah.

So it's not a multifactorial.

Scott Berry: Now, the, the, the
sub phenotypes that you talked

about, the hypo and hyper.

Uh, so a patient that comes in.

So let's talk about the
perspective of a patient.

A patient that comes in is
the, the, the process of this.

So Tony, you see a patient,
they, they fit the a RDS.

Inclusion.

Exclusion are the inclusion
exclusion criteria for simvastatin

and baricitinib identical.

Anthony Gordon: They're very similar, so.

As a platform, we have inclusion
exclusion criteria for the platform.

So as you said, it's essentially
a patient with a RDS.

Um, there are then they're
eligible for the, for the platform.

Scott Berry: Uh huh.

Anthony Gordon: Importantly it is then at
that time they have their blood samples

measured and their assignment to the
hypo or hypo inflammatory sub phenotype.

Done upfront before they're randomized
so that they can then be randomized

essentially into the two strata.

Um, according to the, the sub
phenotypes, the, the inclusion

exclusion criteria for each drug.

Um, as I said, I, I.

Virtually the same, but there have
to be a few exceptions because

the drugs work in different ways.

So the obvious one is if
you're allergic to one of the

medications, you can't have it.

But there are some.

For instance, with simvastatin, you have
to be careful with drug interactions.

So if you're on another
drug, you may be excluded.

There's concerns around patients with
renal failure, for instance, that

would exclude you from Baricitinib.

But essentially they're the same.

And then.

Well those eligibility, once you
know, the sub phenotype has been

determined, the randomization happens
separately within each stratum, the

way you, the patient is allocated.

And so, as Victoria said,
it's then a stratified trial.

Um, so the patients are in
either one of the sub phenotypes.

Scott Berry: so as as a patient comes in,
they're classified by the sub phenotypes.

Right now, I believe
Simvastatin and Baricitinib are.

Uh, um, they're, they're getting patients
from either phenotype when they start.

I imagine you could bring in a
third arm where you already know

or think it would be, uh, uh, not
good to test them in one of them.

And it could, it could, could come
in, in only one, I assume, soon.

Anthony Gordon: It could in theory.

Um, um, and this is an interesting point.

We've had a lot of discussion around
this, uh, particularly actually around

simvastatin, for instance, because.

The evidence from Danny's previous
trial, maybe he should, uh, describe it.

Um, or maybe I should keep it shorter,
that, uh, d Danny ran a, Danny ran a

previous trial, um, of Simvastatin in
a RDS and in the overall population

didn't show any, uh, statistically
significant benefit, but then.

Doing the, uh, sub phenotype analysis
in a postoc way showed there was benefit

in the hyper inflammatory group, but
not in the hypo inflammatory group.

So Danny and I particularly had
the conversation about, well,

let's just go forward in the
hyper inflammatory sub phenotype.

But I think a lot of people want
to see both groups to understand

if there really is a difference.

And I think importantly, if you take
the wider picture drug regulators.

Want to understand the effect I if in
the two sub phenotypes to understand,

um, partly to know should have you
recruited the optimal population, but

also what about safety, for instance,
in the other group, if, if people

started using it outside of a sub
phenotype, what would the effect be?

So we thought at this stage we,
it was important to include them

in both, uh, sub phenotypes.

Um, at, at, at initial.

But whether that changes over time,
I think it will be interesting as

new, um, drugs potentially come in.

But at the moment, yes, we're including
them in both strata and want to analyze

them separately throughout the platform.

Scott Berry: Yep.

Danny McAuley: And I guess just to add
a couple of bits, that the, the patients

were also really pain, that we didn't
miss an effect in the other phenotype

where we, even though we expected
it to work in hyper inflammatory,

and, you know, I think that's one of
the things that we're, we're pain.

That the, the sort of design is
underpinned by, uh, patient and,

and sort of public involvement.

Um, so that, that was actually a,
a, another, uh, key piece of it.

Um,

Yeah,

Scott Berry: Yeah, so, so,
so I think that's great.

And you, you said it's stratified by this,
and that's one of those words that can

mean a whole bunch of different things.

Um, uh, whether it's the randomization,
whether it's the analysis, but it

could be during the course of the trial
and we'll get to the adaptive design.

That you stop enrolling one of the drugs.

So for example, suppose this bears out
that simvastatin doesn't work in hypo,

and you, you could stop enrolling that
in hypo based on the empirical evidence.

And then if a hypo patient comes in,
they would only be eligible for the

control and baricitinib and not, uh,
randomized to simvastatin at that point.

But yet, a, a hyper could be random.

To that.

Um, uh, so you're, you're, you're,
this, this could be enrichment.

We haven't said the term enrichment,
but you know, as you go along, you're,

you're learning this as well, uh, about.

Benefit or not within them That the,
the, the, this is a case where you

don't, what the, let's talk about the
control arm, the common control arm.

You're not using placebos
here, so we're not doing double

blinding or multiple blinding.

So, uh, that pa you're not
blinded to what the treatments

take, the patient is taking here.

And so control is standard
of care, I assume.

Okay.

Okay.

Alright, so we've got the, the this.

A patient comes in, they're
classified by hyper and hypo.

They are then randomized among
the arms that are enrolling, that

always equal across the arms.

Victoria, was there any
thought of doing response?

Adaptive randomization,
uh, in those cases.

Victoria Cornelius: Yeah, I mean,
we talked about this a lot, uh,

during the design and in the
end we decided not to use it.

So

we, we, we did consider it.

And it, and it's interesting actually,
it, part of it was, uh, the, the thought

of around the operational side of these
things, um, as well as only having

sort of the, the two active arms versus

the control as well,
and not being convinced.

Of, of the statistical benefits as well.

So, uh, we, we chose not to
do it, uh, to implement that.

Scott Berry: Yeah, no, I, I think that's,
it's a really interesting question.

If you had six or seven and it was a
pandemic scenario and you wanted to

accelerate drug one to the detriment of
drug two, that might be more compelling.

I think here you're trying to find the
evidence for a drug and accelerating

simvastatin and slowing baricitinib.

Probably doesn't make sense in
what your trial is trying to do.

So I, I it's, and, and you don't
want to add that complexity as you

described for not much benefit.

So I think that makes sense.

Anthony Gordon: Yeah, Scott, just
to add, we, that's exactly it.

As you said, outside of a pandemic
where there is no time urgency, um, this

is about comparing each drug against
a consultant to learn whether it's.

A promising therapy that we
can talk about what happens at

the end of this, uh, platform.

Because this is a phase two platform.

It's to identify promising
drugs that would progress.

It's not about comparing different
drugs, it's, it is working

out the efficacy of each drug.

And so therefore the efficiency of one,
essentially one to one randomization

with the control is I think, um.

This, me saying it to some statisticians
now, I think is the most efficient

design, um, to get your, your answers
quickly for each individual drug.

And so that, that's why we're
keeping it, that fixed ratio.

Danny McAuley: And I think the other
thing, Scott, sorry Scott, the other

thing, Victoria might wanna come in.

You know, the way that we've set up the,
the triggers to stop is that, you know,

you have to, you know, be convincing for
efficacy, but we don't wanna prove harm.

So, you know, we stop relatively
early if the signal isn't promising.

Uh, almost trying to back
the winner type approach.

Scott Berry: Okay, so we've talked
about what it looks like for a patient

going in now when an arm comes in.

So you have, you have a master
protocol where when you bring

Simvastatin or Baricitinib or arm
C in, it comes in as an appendix.

Let's talk about what that looks like
from the perspective of an arm and maybe

even thinking about a pharmaceutical
company that wants to put an agent in and

thinking about, okay, what happens to my.

My, my, my baby as it
goes into this trial.

What is the, the design for that?

So the, the, the arm enters in
and it wants to enroll in both,

uh, phenotypes, sub phenotypes.

What is the design for an arm?

Victoria, what does that look like?

Sample size,

Victoria Cornelius: Yeah.

So

Scott Berry: analysis.

Yep.

Victoria Cornelius: yeah.

So, um, this is, again, this is
something we've been, uh, obviously

spending a lot of time on the
current design with the arms.

And actually one thing of the things
I did want to talk about, uh, with,

um, this design, because we are,
we are, we are really lucky to be

doing this stratified trial, this

precision medicine approach is
actually, in order to do that,

we've got to have devices.

And so one of the design considerations
for the whole platform was that

the number of devices we can get,
because that's gonna restrict the

number of centers we can include,
is gonna restrict the sample size.

So that was So we've

Scott Berry: So sorry, can
I, devices meaning ability

to identify hypo and hyper.

Victoria Cornelius: Yes.

Scott Berry: Uh, okay.

Okay.

Victoria Cornelius: so one of the very
early design considerations is around,

around, uh, getting enough evidence
for the phase two, doing that robustly,

as Danny had said, but actually just
getting enough evidence such that

it, it's good enough to get through.

Right.

Let's test this in phase three.

So, So, we, so there is a sample
size consideration there, so regard

to bringing in an additional arm.

I think part of that will be, have
to come within the whole framework.

Of, uh, how far we are along with, um,
the current two active interventions.

Yeah.

Scott Berry: So, uh, go ahead.

Victoria Cornelius: yeah, Sorry.

just to say it, so at, maybe if it
came in now at the start, it would

have very similar considerations.

Um, but uh, later on we would have
to redo and relook at those things.

Scott Berry: Okay, so a a, the, the
design for Simvastatin, for example,

is it comes in and it's analyzed
separately in hyper and hypo.

Uh, and maybe we'll come back to the
question of, uh, borrowing or, or

independent analysis within there, but
you've got an adaptive design that every.

Every three months, you do an interim
analysis, and when a minimum sample

size is reached for that particular
arm, it enters into the possibility

for stopping rules where it could be
stopped for promise, or it could be

stopped for lack of promise, presumably,
or it could hit a maximum sample size.

So if my arm comes in,
it goes through these.

Quarterly analyses and, uh, the
maximum sample size, just roughly,

uh, for, for an arm is what?

Victoria Cornelius: It's
around about 505, I think.

5, 2 5.

For each arm?

Scott Berry: For, for each sub phenotype?

Victoria Cornelius: Yeah, for each arm.

Scott Berry: Okay.

So once it reaches that max,
where, where we've got enough

information, this is phase two,
but it could stop sooner than that.

And it can stop at different
times within hyper and hypo.

And for different reasons, it could
be promising and hyper simvastatin

if, if it reproduces a previous
result and it could lack promise in

hypo, uh, as a potential within that.

Now, the, the analysis that done, uh, the,
there's, there's a huge amount of question

in these, in intensive care type trials.

There's what's the endpoint?

Um, so what is the primary endpoint
that is used for the determination of

promise or not promise for an agent?

Victoria Cornelius: Yeah,
a, uh, a, a great question.

Um, so we had, uh, again, another
thing we talked about a lot, um,

obviously we are trying to, the, the
obvious one would be to use mortality.

Um, but we were trying to, we were
trying to move beyond mortality

in these situations because.

We want something, um, partly because the
statistical considerations around using

it, which we need thousands and thousands,
uh, but also trying to get an, uh, sort of

capture an important part of the patient
journey as well within the outcome.

So terribly inspired
by the REMAP cap trial.

Um, we have got the primary
outcome of 28 days of a composite

the same, but it's for 28 days.

Organ free support.

Um, so we've got our mortality being
our minus one score and then, and

we've got it up to zero, up to 28 days.

Organ free support being the best outcome.

Scott Berry: Okay, so you described these
ordinal outcomes where death is the worst

outcome, and then being on organ support
for 28 days is the next worst, and so on.

And so you've got this ordinal
outcome which you analyze

using a proportional odds.

Bayesian proportional odds model.

Yep.

Victoria Cornelius: Yeah, that's right.

Yeah.

Scott Berry: Okay.

And, um, I, and in a really interesting
question, and I think it almost becomes

more interesting, part of what I found
intriguing about the trial is you've got

an appendix for subphenotypes where the
science may evolve, where you learn that

there are really important subphenotypes
and you add them to the trial.

Right now you have two, and so you
have this question when you analyze.

Um, uh, hyper and hypo.

You run a completely separate model.

Within each of these, and
so there's no borrowing.

So if a drug is doing well in one
and well in the other, there's no

kind of borrowing strength there.

Partially you're worried about
heterogeneity, and that may

have the opposite effect.

This could be a bigger problem
if you add another sub phenotype

to the trial that the question is
then how do we analyze across this?

Was there any discussion about
potentially doing borrowing

Victoria Cornelius: There, there was a
lot of discussion about doing borrowing.

Um, philosophically I
have a problem with that.

Like you say, we're actually out and we've
got quite strong, you know, we've got some

underpinning evidence we're expecting.

Differential responses here.

So it feels, the premise of it doesn't
feel correct and, uh, to, to be borrowing

across something where we don't, uh,
expect it, you know, and we wouldn't, we

would only then borrow if it was similar.

And then I actually, I, as
a statistician, I struggle.

With only borrowing when
it's similar as well.

So I, I do struggle a little bit with
the borrowing concept when it's not

underpinned by other justifications.

Yeah, so we did discuss it.

I think ultimately it was because,
uh, we are expecting differential

treatment effects here, so we wouldn't.

Scott Berry: Yeah.

Yeah.

The, the problem of dimensionality could.

Be an enemy to you though right now
you've got maximums of about 502 groups.

Suppose you added another sub
phenotype that made it two by two.

Now 504 different groups becomes 2000.

If you had two by two by two,
now you've got, uh, you know,

you, you're making this.

8,000, you're making
this an enormous trial.

4,000, an enormous trial
without some level of borrowing,

but you're not there yet.

You're two where, and, and borrowing
in two is always kind of a weird thing.

Yes or no?

Uh, I, I, I, the, the science of
this, do you expect to have more sub

phenotypes in the next few years?

Tony?

Anthony Gordon: Yeah, it is a
really interesting point, Scott.

Um, I think there is, there
is a strong possibility we

might, um, Victoria might get.

Try and urge just

Scott Berry: Yeah.

Anthony Gordon: to hold back our
enthusiasm, but as Danny alluded to right

at the beginning here, the, the, there
is this, the clinical, uh, syndromes.

We, we, we talked about a RDS, but there's
a massive overlap with sepsis in general.

Um, you know, ar they're like
two big parts of a Venn diagram.

They overlap, but actually.

They're very similar,
the biological processes.

And so you could imagine we might want
to include a broader population with

sepsis and a RDS because the underlying
biological processes are similar.

And also the, these hypo and
hyper inflammatory phenotypes are

probably the most well described.

But there are others
work we've been doing.

Um, with colleagues in Oxford, for
instance, uh, Julian Knight, um,

around gene expression profiles that
is looking really promising as well as

identifying, um, sub phenotypes that
will respond to different treatments if

we were then to bring them all together.

Then I think you're right.

Making a good point of why we might
need to consider borrowing, but

I, I don't think we're there yet,
but we're always thinking big,

um, and looking to the future.

But I think probably that's where we need
to go to take the whole area of critical

care research, um, further forward.

Um, but we, we might need to
come and revisit that, um, and

do some brainstorming about what
that actually would look like.

I think.

Scott Berry: Mm.

Danny McAuley: And I guess Scott, the
other thing to say, you know, we're,

we're saying we've got this heterogeneous
mass and we've divided into two, and

the idea that two is the ground truth.

So I think this is just the start of
the, the journey to get, you know, better

definition of all of these sub phenotypes.

And that's one of the really
important other design features were.

Capturing biological samples
within the trial as we go along

to hopefully understand the,
the, the syndromes better as well

Scott Berry: So I imagine,
suppose Simvastatin enrolls 500

and 500 and it gets to the end.

You've got these

Danny McAuley: and works and
and, obviously he's gonna work.

Scott Berry: right.

Yeah.

Well that was clear.

Um, but the, you've got the sub
phenotypes, but I imagine you're

gonna do a great deal of after the
fact investigation of heterogeneity

treatment effect over all kinds of
different, uh, other things when the

data are finally in, uh, also creation
of new sub phenotypes based on the

data that's accruing within this trial.

Yeah.

Anthony Gordon: Yeah.

And, and Scott, I think
that's an important part.

These, these phenotypes have really
emanated from postdoc analysis of a

more standard traditional A versus B.

Trial.

And I think part of the international
collaboration here is about

trying to bring those who are
interested in this field together.

And this isn't just clinical trial,
this is, there's people in the

international group that do much
more of the basic science and will do

that biomarker work, that discovery,
and actually try and take the whole

field, uh, forward, uh, together.

And so, rather than.

Do it on an ad hoc basis, can we sort
of plan it upfront that we are actually

setting up the whole infrastructure,
randomized patients in the trial, learn

about those specific questions that
are the randomized questions, and at

the same time, um, build that precious
resource that, um, is exploratory but

drives the next question that can come
into the platform, um, in, in the future.

Scott Berry: Yeah, it makes a ton of sense
to have this central resource of data in

a RDS with multiple treatments to do this
further, moving the science forward of

understanding a RDS and endpoints and, uh,
uh, different phenotypes is, is amazing.

Uh, where if everybody did
these trials separately.

E, this is a much weaker.

Global effort in a RDS, which is one of
the, the, the things that I don't think

is talked about much, um, is that the,
having all this data together, so this,

we, we've talked about the huge advantages
of this, this bringing all this data

together, bringing the multiple arms in.

Uh, what are the challenges,
uh, uh, in this moving forward?

And I, I assume some of them
are operational, logistical.

I, I don't know if they're funding.

What are the challenges that you're
going to be facing, uh, maybe

initially in running this trial?

Victoria?

What keeps you up at night,
other than having me on your

DSMB, what keeps you up at night?

Victoria Cornelius: I, I definitely wanna
bring in to Tony and DNI on this as well,

but just so as you introduced me at the
start, I also direct the, the kind of.

Trials unit at Imperial.

And actually for us, this is a
really important, uh, platform

'cause it's our first platform.

And, uh, we've got, you know, we've
got a lot of trials in our books and

we've got about IET at the moment,
but this is our first platform.

And so we've had to learn quite a
lot, uh, uh, across the whole team.

Obviously the.

Statistical team.

Um, uh, it is, it's, you know, the
funding for that, how to deliver and

work together, all the rest of it.

Our standard operating procedures.

We've got our operational side and also
our, our clinical data systems as well.

So it's been, it's been quite a journey
for us o on that side of things.

And like I say, uh, we talked about the
funder, but actually NHR uh, funded us.

For an accelerator to get ourselves ready.

So we had a year, uh, of preparing,
you know, designing that trial.

And then we've had a very
generous funding, uh, to get the

platform up and start as well.

So I think we wouldn't have been able to
do that now that we're embarking on it.

Um, I'm gonna, I'll, I'll let
Danny and Tony come in to what,

what may be worrying them.

Danny McAuley: So I, I mean, I,
I guess this is probably a, um,

a global issue to some degree.

But, but, you know, clearly we want to
protect patient safety at all costs,

but the actual process and governance
process of getting research established

particularly, um, in our experience
in the UK is very, very difficult.

So I, I think, uh, that that is
the one thing that that keeps.

Well, not one, the one thing, but
one of the many things that keeps

me, uh, up at night and in terms of
how we actually embed research more

efficiently and, and increase, uh,
recruitment into these important trials.

Under just the bureaucracy that we
face in, in getting the trials open.

And it's a really important, uh,
sort of national priority, I guess.

The, the, the, the one other thing
we, we work very closely and we

talked about, um, the device that we,
uh, use, uh, to phenotype patients.

That's on a, a collaboration with,
uh, a company called Randox and,

uh, getting the, uh, sort of, um.

Devices deployed at sites despite Rand's.

You know, full support has
been a, a challenge as well.

And, and scaling that up
internationally then, um, presents,

uh, a potential, uh, challenge.

I think, we'll, we'll overcome
it, but they would be the, the

two that worry me the most.

Uh, but let's see what Tony has to say.

Anthony Gordon: Yeah.

Um, well now that we are funded,
obviously that was the first issue

is getting, getting the money and
also particularly getting money

in more than one, uh, region, you
know, the international trials.

But we, we, we are making
good progress with that.

Um, I think.

Obviously there's elements of this,
as alluded to, it involves both the

device to measure the sub phenotype,
the, the biomarkers and the drug.

So you've got, you've got both
device and drug aspects to manage.

It's large and it's international.

So I, I think the coordination
of that by the trial management

team is so crucial and I think.

Uh, it doesn't keep me up at night 'cause
I know I've got a good team that do it.

But, um, I just would, I, I
think I just need to recognize,

uh, those, uh, team members.

Uh, so in this case our, our team leader,
Janice Best Lane, just as phenomenal.

Her and colleagues have,
uh, learned through COVID.

They worked, you know, on
the, um, in remap Cap and the

challenges there, but how you keep.

The whole, um, ship, um,
sailing, um, to manage this,

you and you do need that team.

And so I think, uh, sort of advice
to anybody setting out on these sort

of things is you have to have that
support team there to manage all

aspects of the protocol that the modular
nature of the, um, of that protocol.

We are moving to try and do it all
electronically rather than on paper,

for instance, that things like that
help, but then to coordinate it and

keep all the sites involved around the
world, um, to coordinate when there

are so many things that can change.

I think running any international
trial is a challenge, but if it's a

fixed protocol that doesn't change.

It's still a challenge, but
if you're gonna change it over

time, how you manage that.

And so, um.

I, I think it can be done,
but you need experience.

You need an experience team.

And so I just want to recognize the hard
work, uh, they put into to deliver it.

Danny McAuley: I just wanna second that.

I mean, Janice and, uh, the team
managing me, Tony and Victoria

in itself is a challenge.

Nevermind the, the, the whole trial
infrastructure, but the, the, the

other thing I just wanted to pick
up on was that the funder, and it

is this problem that we've had in
the past of this idea of going to

multiple jurisdictions and facing.

Double, triple, quadruple
jeopardy to get your trial funded.

And I, I think we're starting
to see funders work together,

uh, on big ticket items.

So I think hopefully going
forward, uh, that might be a, a,

an easier issue to get sort of,

Scott Berry: Hmm.

So there,

there's a potential moving forward that
you could bring in a pharma sponsor,

uh, that, that could bring in funding.

Another, uh, a way to
bring in funding of this.

What does the trial
look like going forward?

You know, what does
steady state look like?

What would be the optimal number of arms?

And to some extent, there's a risk that
one of these two treatments say drop in

hypo, the other one drops in hyper and
you're now running kind of an AB trial

and each, you only have one arm in that.

Would you like to have.

Five arms, two arms.

What is steady state and
what, what is moving forward?

What would you like this to look like?

Tony?

Anthony Gordon: Um, so importantly we
have a pipeline, and that was something

we set up right from the beginning.

So there are already drugs in the,
um, in the queue waiting to come in.

So if, if we are lucky
and get an early win.

And that's an answer.

I mean, that if, if we learn it
doesn't work, that's still a win.

We, we know it doesn't go forward and
then we've got drugs ready to bring in.

So, and importantly I think we have
a process by doing that as well.

So, um, and part of that is internal.

We've have a prioritization committee,
but we've also got an external committee

to advise us, um, that includes clinical
and pharmacology expertise to guide us.

And I think that's something I've learned.

From platforms, how you
decide what comes in.

It's good to have that
independent oversight.

I think it depends how
many sites we have open.

Um, and Victoria alluded to this, if
we, we were talking 70 sites initially,

we think the three arms with the two
active interventions is probably the

right number to get answers quickly.

But if we manage to open 200
sites, um, because we are able to.

Measure the biomarkers in, you know, more
laboratories, for instance, I think we

could then bring in a, you know, a third
active intervention at the same time.

Um, it it, so yeah, with that
extra resource, we could do

more, but we, we thought sensible
to, you know, manage that.

Um, at least initially.

Scott Berry: Mm.

So Tony, suppose you get word back.

Uh, so, sorry, sorry.

Uh, Danny, suppose you get word back that
Simvastatin is successful in this trial.

It's promising.

What happens?

Uh, do you run a phase three trial?

Does Panther run a phase three trial?

It's, I, I think this is completely,
uh, off patent and generic.

What happens?

Danny McAuley: So I, I guess simvastatin
is, um, probably a wee bit different from

maybe some of the other drugs coming in.

And I'll maybe explain,
picking up on Tony's point.

You know, there probably is a, a pretty
strong prior in the hyper inflammatory

group that, uh, simvastatin might work.

And if we showed in a big phase two.

With, you know, good certain day of, uh,
um, positive effect that might be enough

to change practice in that, I think fairly
unique situation in most situations.

You know, taking, uh, baricitinib,
there is some proof of concept

largely from, uh, COVID, but it's not
probably enough to change practice

even in the setting of a positive
readout from the, uh, Panther trial.

So to me, in that setting,

that's where we would.

We would feed the beast that is, uh,
remap cap, which is so well set up to

answer these multifactorial questions.

And that's other, that's one of the
other things that we, you know, talked

about in terms of pandemic preparedness.

You know, one of the nice things
about Panther and Remap Cap working

closely together is that, you
know, if we did a pivot in pandemic

preparedness, we would be able to.

Quickly prioritize phase two
assets that then could go into

the, the, the phase three that
that remap cap does, uh, very well.

So that's how we sort of,
uh, imagine that would work.

Scott Berry: Mm.

Anthony Gordon: Scott, can I, uh,
suggest another alternative as

well, and it's the sort of industry.

Uh, perspective here.

So I think that, you know, what we
think is that, um, pharma companies

will have a number of drugs already.

Maybe they're already, um, got
licenses for other respiratory

or other inflammatory conditions.

It's a big ask for them to, um,
maybe run their own phase 2 trial.

That, you know, we, as we've talked
about, the issues with critical

care trials, um, that haven't.

Produced, uh, successful results.

The idea would be we have a platform
already up and running their asset that

they have maybe using in a chronic lung
disease, put it into the Panther platform.

We've already got the
infrastructure there.

We've essentially de-risked it.

They learn whether the
drug is looking, promising.

If it is, they get that result and
they might choose to do their own

licensing trial, um, for this indication.

Or they learn it wasn't effective,
and they learn that, uh, quickly

and we've sort of de-risked
the whole process for them.

So we think this would appeal
to them and, and gives a better

chance for us to actually, uh,
bring new therapies to patients.

And it's that academic industry
collaboration that, um, I

think could be really helpful.

Scott Berry: Uh, yeah.

And the amazing thing is for, for that
pharma company to go out and run that

phase 2 trial and go get 70 sites and,
uh, get devices and all that might, might

cost x million dollars for you guys to do.

It might be a 10th of that.

Uh, and the cost, I, and I don't know
what it is, but some fraction of that

to, to do this, to take this shot on
goal to, to the benefit of all, maybe

you guys get 10 times as many shots.

Because, because of that, it's
just such a, an awesome potential.

It seems by the way, that many of
these platforms have come about by

patient organizations in pancreatic
cancer, in glioblastoma, in a LS where

not things don't work and companies
stop taking shots on goal because

it's so challenging and many of these
platforms are developed for Exactly.

Trying to solve this, the,
the, this, this dilemma.

It's, it's a huge potential.

Danny McAuley: Y Yeah, no,
I think that's right Scott.

And, and as Tony said, the, the
idea of de-risking in a high risk.

Environment that is critical care.

You know, and the sort of the, the,
almost the taking smart risk, uh, is, is

sort of the mantra that we're trying to
sort of get across to, uh, pharma, that

that will be of benefit to, to them,
but also to, uh, patient, uh, outcomes.

Scott Berry: Hmm.

And, and, uh, uh, a hugely
positive thing with a successful

drug in a RDS that improves.

Organ support free days of the,
the, the size of the potential,

uh, win and the, the benefit.

Uh, so the huge potential at
the end of the day just been a,

a challenging area to go into.

All right.

Uh, a very cool trial.

I'm, I'm, I'm proud to, to watch this
go on and excited to see how this goes.

Uh, and lots of times I will be in
the interim and, uh, I appreciate

you all coming here in the
interim, uh, and joining us all.

So thank you all.

Victoria Cornelius: Thank you very

much.

Thanks.

Anthony Gordon: Thanks.

It's been brilliant to chat about it, sco.

Cheers.

Scott Berry: Yeah.

Wonderful.

Thank you.

And everybody else, we will
be here in the interim.