Latest / Cardio Pod: Heart of the Matter / Should Investigators Have Access to Trial Data?

Transcript

Should Investigators Have Access to Trial Data? · Cardio Pod: Heart of the Matter

0:06
Well, good afternoon, everyone. I'm Chris O 'Connor.
0:08
I'm the president of the NovaShare Heart and
0:11
Vascular and principal investigator of the Heart
0:13
Failure Collaboratory. And we're here today with
0:16
Heart of the Matter, our podcast discussing various
0:20
issues that have come up through the Collaboratory's
0:23
efforts throughout the year. And today's topic
0:26
at hand with our distinguished panel is, should
0:29
investigators have access to clinical trial data?
0:33
And I'm pleased to be joined by distinguished
0:35
panel, Dr. Robert Califf from Durham, North Carolina,
0:39
Dr. Marvin Konstam from Boston, Massachusetts,
0:43
Dr. Mike Bristow from Denver, Colorado, Dr. Scott
0:47
Salmon from Boston, Massachusetts, and Dr. Mona
0:51
Fouzad from Arlington, Northern Virginia. Well,
0:56
welcome, team. And I want to get us started by
0:59
asking Scott. Why is this an issue? Why is this
1:05
an issue that should investigators have access
1:08
to clinical trial data? So first, thanks for
1:13
having us on and doing this, Chris. And I think
1:18
it's an issue that many of us have struggled
1:21
with for many years. Those of us who've done
1:23
clinical trials working with industry. And these
1:30
aren't trials that we're doing as investigator
1:32
trials only, but ones that we're doing in collaboration
1:35
with industry. We often at the end of the study
1:39
get shown data. We get shown tables, figures.
1:43
But how many investigators truly have direct
1:48
access to the data? And I think that it is important
1:53
that investigators be able to vouch for the data
1:56
when they write that paper. that goes to The
2:00
Lancet or The New England Journal or JAMA, and
2:03
they have to say, we vouch for the integrity
2:06
of the data. What does that mean? So it's become,
2:10
I think, a bit of an issue because there are
2:13
sponsors for whom they expect that the investigators
2:18
will work with them on the data, have access
2:22
to it, do statistical analyses and so forth.
2:25
And there are also... companies that are much
2:28
more in control of the data and provide tables
2:34
and figures potentially. And maybe that's good
2:37
enough. And I guess that's really the question.
2:40
Do we as academic investigators, as partners
2:45
with industry, necessarily need direct access
2:50
to data to be able to say? And vouch for the
2:53
say that the data was was assessed correctly
2:57
and vouch for the data in the manuscripts that
3:00
we write. I think we do. And I think that investigators
3:03
should have access to directly to data. But it's
3:09
worth a discussion. And there are obviously different
3:12
models. I started in this business many years
3:18
ago working with Mark Pfeffer. and Rob Califf
3:22
on a trial called Valiant, where it really was
3:25
a true industry -academic partnership, and that's
3:28
the model that I thought worked incredibly well.
3:32
And I wish that happened more often, but often
3:36
investigators are not given direct access to
3:40
data, and it's, I think, important. And we learned
3:44
a few years ago that when some investigators...
3:50
didn't have direct access to industry -supplied
3:53
data. There were erroneous publications. This
3:58
was during COVID that resulted in at least one
4:01
major journal, The Lancet, putting out a statement
4:05
saying that in every instance where an academic
4:08
investigator is working with industry, where
4:10
there's an industry -academic collaboration,
4:13
that at least one academic investigator has to
4:16
have direct access. to the data and be able to
4:20
say that. I think that should be a norm, but
4:24
I'm really interested to hear what everybody
4:25
else thinks. Well, Rob, you've had the most unique
4:31
position to be at every angle of this question.
4:36
You were the founding director of the Duke Clinical
4:39
Research Institute. You've been On the sponsor
4:42
side at Verily Alphabet, you've obviously FDA
4:46
commissioner, two terms, and you've been a journal
4:49
editor, the founding editor of the American Heart
4:52
Journal. Well, not founding, but one of the editors
4:55
with Dan Mark of the American Heart Journal.
4:57
What do you say about this? Is this, you agree
5:01
with Scott? You think it should go further? You
5:04
think it's too far? Well, let me start by trying
5:07
to refine the question. We'll see if others agree.
5:10
When you say investigators having access, let's
5:13
say we're doing a 2 ,000 -person trial involving
5:16
50 sites. I don't believe you're talking about
5:19
every investigator at every site having free
5:22
reign on the database to do whatever they feel
5:24
like doing. You're really, I think, mostly talking
5:27
about in this. Few cases, and I'm going to come
5:32
back to this point, a few cases where there is
5:35
an academic organization and coordinating center
5:38
involved. And again, I use the word academic,
5:41
really not necessary to mean university associated,
5:45
but a group of interested clinical people that
5:49
are running the trial where they have that capability.
5:53
The question is, should they only get tables
5:56
and output from the sponsor or should they do
5:59
their own? with the ability to look at the raw
6:03
data. Is that the question that you're... That's
6:07
it. That's correct. Okay. So first, I have to
6:10
make a couple of quick rhetorical points. First,
6:12
we always keep using the word industry to describe
6:14
the medical products industry. The healthcare
6:17
delivery industry and the big behemoths like
6:21
MGB that Dr. Solomon is associated with are much
6:26
larger than the medical products industry. And
6:30
if you don't think that they care about what's
6:33
said in these publications or have a tendency
6:36
to gatekeep on what questions are asked, you
6:38
really ought to look at it more closely. So I
6:41
sort of object to the word industry and academia,
6:43
like academics are Puritan white knights who
6:47
do everything right, and industry is greedy and
6:50
only looking out for itself. I'd also point out
6:54
that when it came to COVID, the most egregious
6:56
offenses were by academics who were publishing
6:59
willy -nilly ridiculous observational studies,
7:03
making claims that were just completely incorrect.
7:06
And over, I think it's like 98 % of published
7:10
COVID studies were SCTs, small crappy trials
7:14
that really had no chance of answering the question.
7:19
So the way I look at it, but let me also on the
7:21
other side, let me just say this is not a new
7:23
problem. And when Dr. Solomon was in junior high
7:26
school, I think we were doing the gusto trials.
7:29
And I would say early in my career, the majority
7:33
of trials where Duke would be the coordinating
7:36
center, we were not allowed to do because the
7:38
industry would not allow access to the database.
7:42
And that was a requirement of the university.
7:46
And so that alleviated somewhat when we got clinicaltrials
7:50
.gov through and did away with the ridiculous
7:52
notion that you should be allowed to do human
7:55
experiments without letting anyone know publicly
7:58
that you were doing them. And as predicted by
8:02
the industry leaders, the opposite happened.
8:07
They said this will destroy the business model.
8:09
It really didn't. And so I would say since the
8:12
results are going to be publicly available anyway,
8:15
it shouldn't be that much of an argument. But
8:19
I think what is in play is mutual distrust about
8:22
the motives of each side. And I think the concerns
8:25
of academia is that industry is bound in a fiduciary
8:30
sense to develop their product as an asset in
8:34
the case where industry funds a clinical trial.
8:36
Industry is suspicious that academics become
8:39
famous by causing trouble and pointing out things
8:42
that, you know, aren't so good. And so this all
8:47
turns in, in my view, into a social relationship.
8:51
And I would say on balance, qualified, reputable
8:55
academics with a track record should have access
8:58
to the database. But turning it loose to anyone
9:02
who wants to look at it, I think has a number
9:05
of problems associated with it. So terrific points.
9:10
And I want to get to a point of. What about negative
9:15
or neutral trials where there's no longer any
9:17
interest by the sponsor? But Mike Bristow, you've
9:20
been on both sides of the coin also. What's your
9:24
take on this discussion? Well, yeah, so I think
9:30
there are levels. Excuse me. But I think there
9:35
are levels here that we need to talk about in
9:38
terms of investigators. And I think we have,
9:41
you know, we're talking about. multiple domains
9:46
here as well in terms of responsibility or fiduciary
9:50
duty. You know, the standard investigator in
9:55
a multicenter trial, I don't believe, expects
10:00
or should have access to the raw database because
10:05
they're probably not going to do anything with
10:06
it for one thing. On the other hand, there needs
10:10
to be some fiduciary duty. Among the people running
10:14
the trial are really driving the trial, meaning
10:18
steering committee, for example, or people who
10:22
are chairing sub -studies. The chairman or overall
10:27
PIs of the study are typically on a steering
10:31
committee anyway. There may be other people,
10:34
though, like sub -studies chairman that need
10:37
to have access to the raw data or certainly could
10:40
get it if they wanted. And then, of course, you
10:44
know, we have the famous DSMV we've been talking
10:47
about on other podcasts and meetings. They get
10:53
to see the data for sure and ensure some aspect
10:57
of legitimacy to it. So I think it's not just
11:02
an issue of blanket investigator. There are various
11:04
people in different roles and we do have different
11:07
responsibilities and should have them. And I
11:11
think another thing is that the way to deal with
11:14
this is up front in the CDA or a contract that
11:18
investigators have with the sponsor, what's expected
11:22
of them and what they expect should be in that
11:25
up front. So there's no misunderstanding at the
11:28
end of the trial. So I made a bunch of generalities
11:32
here that are not very useful, perhaps, but this
11:35
is more complicated than just an answer to the
11:38
question that... was posed at the start of the
11:41
podcast. Well, I think that brings up good points.
11:46
And Marv, I want you to jump in here. I think
11:49
Mike brought up the issue of the DSMB. Could
11:52
that be a safety net if a sponsor says, no, I
11:56
don't want the investigators to have access?
12:00
Yeah. Well, I've been on a lot of DSMBs and I
12:04
guess one thing I agree completely. With Mike,
12:08
this is pretty complicated, and there are a lot
12:10
of different levels to this. It really depends
12:12
exactly what we're talking about. TSMB is an
12:15
interesting example. So does the DMC, do the
12:19
members of the DMC, does the chair of the DMC
12:22
actually have direct access to the data files?
12:26
And the answer is usually no. There's usually
12:29
a statistician or a statistical group, you know,
12:32
that manages those and are therefore unblinded.
12:36
And we tend to have a lot of faith in them. But
12:40
I haven't been in any data monitoring committee
12:42
where the chair stood up and say, I want you
12:45
to transfer the data files to me. So that's sort
12:48
of another level. And it brings up the issue
12:53
of what do we mean by the data? And for example.
13:01
Does the steering committee or executive committee
13:04
want to see the clinical trial report, clinical
13:08
study report, the CSR that's provided to the
13:11
FDA, et cetera, which is a listing of all of
13:17
every analysis that's been done. Of course, there
13:20
are a lot of analyses that haven't been done.
13:21
Or are we talking about actually transferring
13:24
the data? And I think, you know, both questions
13:27
are in play. And if you're going to take the
13:30
actual data files, of course, the question, what
13:33
are you going to do with them? And so, you know,
13:36
you have to either personally be able to manage
13:39
that and analyze that from a statistical viewpoint,
13:44
et cetera, or you have to be working with somebody
13:46
who does. Otherwise, it's not of any value to
13:50
you. I want to get one thing off the table because,
13:53
you know, the issue of What if an investigator
13:59
wants the data from their site? Is that something
14:05
that should be allowed? And Scott's just shaking
14:08
his head, and I would shake my head vehemently
14:10
as well, because really the value of this database
14:15
is the aggregate nature of it, the fact that
14:21
there's so many patients involved, the strength
14:23
of that, and to have somebody go up and publish.
14:27
their own data, I think could really undermine
14:29
that. So let's get that off the table. But I
14:34
think that otherwise the authors of the paper
14:38
clearly need direct access to the data. And I
14:43
would say that certainly the chair of the writing
14:49
committee or steering committee or executive
14:52
committee, you know, should have that. And when
14:57
you get, and I think the entire writing committee
15:03
needs to verify that, yes, I've had access to
15:07
the data and I could do with it what I want or
15:09
get the analyses that I want. I may differ a
15:13
little bit around the investigators out there.
15:16
I certainly don't think you can distribute the
15:20
data to 500 investigators. There may be somebody.
15:25
who actually has the ability to analyze data
15:28
and wants to have it for academic purposes or
15:34
whatever, I'd be in favor of that happening,
15:37
but that rules be written about what that means,
15:41
who should have access to that and who shouldn't.
15:48
If somebody does that, should the original writing
15:53
committee or steering committee, have a stay
15:56
in that paper or participate in that paper. A
15:59
lot of questions that come out of that. But I
16:02
think somehow or other, if somebody like that
16:04
comes along at some appropriate point in time
16:07
and says that, I think we should consider access
16:10
to them. So complicated questions. I spent the
16:14
whole week trying to think of a reason why investigators
16:18
should not get the data. And I absolutely couldn't
16:21
come up with one. So I guess I'm generally in
16:23
favor of it. Chris, I wonder if this is really
16:27
in many ways a preamble to a broader question.
16:30
And I think, Rob, you hinted on this, which is
16:33
really, what is the nature of the partnership
16:36
between industry and or the broad industry, as
16:43
you'd call it, and the academic investigators
16:50
who are really working with them as partners
16:54
in a particular study. And I don't mean the broad
16:57
group of hundreds of investigators out there,
17:00
but more what we would typically call the executive
17:03
committee or a steering committee on a trial.
17:06
And there's no guidance for this whatsoever.
17:09
I mean, we've all done it in lots of different
17:12
ways and there are different models. And in some
17:15
cases, that group has some fiduciary control.
17:20
In some cases, they have no fiduciary control
17:23
and are simply there as advisors. And I've seen
17:28
situations where they were specifically not called
17:33
a scaring committee so that the company did not
17:36
have to necessarily defer to them with any decisions.
17:42
And in other cases, in many cases, we're asked
17:47
when we publish papers, Who's making that decision?
17:51
And you have to be able to say generally it's
17:55
the whole group of authors who've made that decision.
17:59
I think that guidance along these lines would
18:04
be enormously helpful and serve as a template
18:07
for future industry academic partnerships. Our
18:11
institution, probably just like Duke, will not
18:15
accept a situation where A company has the rights
18:20
of review. I'm sorry, rights of approval. They'll
18:25
only accept rights of review. But I have at least
18:30
on one occasion had to walk away from a very
18:33
large project for that reason because we couldn't
18:37
come to terms with the company. They wouldn't
18:39
give us publication rights or rights to review.
18:41
I think this could be enforced by journals potentially.
18:46
And also, we, groups like the Heart Flare Collaboratory,
18:50
could potentially provide guidance for these
18:53
kinds of relationships. And I think you're right.
18:57
I mean, Dr. Fusada, from the journal standpoint,
19:00
we've seen much more instructions from the journals,
19:04
the journals requiring that data transparency
19:08
and access is available to the investigators
19:12
as part of the data sharing agreement. Could
19:15
the collaboratory take a position, put together
19:19
a position statement to help standardize this
19:23
across trials? What we see now is that, yes,
19:26
there are best practices, like Scott says, like
19:29
Rob, all the investigators here, but there is
19:32
variability. And particularly when you have a
19:38
negative or neutral trial. We want that data
19:41
to be out there. We want people to move that
19:45
data along. Often the sponsors don't want to
19:47
do that anymore. And the question is, I think
19:52
that social contract that you've talked about
19:54
at the beginning would be valuable. But Dr. Fuso,
19:57
what are your thoughts? Yeah, I agree. I mean,
20:00
I've also been on several different sides of
20:04
this industry, FDA, journal, academic. And I
20:09
agree with everything that's been said. I think
20:12
there is a lot of trust between the parties that
20:16
has to occur. We definitely at the Journal saw
20:20
investigators with access to data that was clearly
20:25
not being understood correctly. And so I think
20:29
that's a really big issue. I think some of these
20:32
trials have a lot of nuances that need to be
20:34
understood. really the senior leaders are the
20:38
only ones who really understand them. And so,
20:41
you know, for a journal editor, it does become
20:44
very problematic to see papers with different
20:48
results or different interpretations of data
20:53
that are coming through. So I agree with everything
20:57
that's been said. I think, you know, I was thinking
21:00
as you're talking, maybe one of the changes that
21:04
need to be made is the question that Scott started
21:06
with, which is, I know that the journals ask
21:09
the lead author to take responsibility for all
21:15
of the content of the paper, but maybe the person
21:18
who's done the analysis should also be vouching
21:21
for the portion that the other investigators
21:23
can't. Rob, give us the last word. What can the
21:29
Collaboratory do here to standardize this, move
21:33
the field forward, reduce the variation in what
21:37
we see so that we have better transparency and
21:40
greater access. I think you're good. I'm muted
21:48
here. I would sure encourage the laboratory to
21:50
put something together that represented a consensus
21:53
on best practice, recognizing that... For some
21:58
of this, best practice just means what people
22:00
think is better. There's no empirical proof of
22:03
what's best. But I'd also point out, as we go
22:07
forward, there's going to be a lot to be learned.
22:09
Remember that the NIH now has a policy to make
22:12
the raw data available. So if an NIH -funded
22:15
study, and I know, Scott, Dr. Bhattacharya is
22:19
Harvard's best friend right now, but in any case.
22:24
He's promising to even take this a step further
22:28
in terms of the transparency of clinical trials
22:31
data. So that's going to provide a model where
22:33
a lot of this can be tested. On the other hand,
22:37
one thing I was reminded of at FDA is that 95
22:40
% of clinical trials don't involve multiple sites,
22:44
don't have an academic steering committee. They're
22:47
run by the sponsor as a single site academic
22:51
medical center trial. None of this really is
22:54
pertinent to that. And so I think the question
22:57
of when that structure is optimal is another
23:00
thing the Heart Failure Collaboratory could opine
23:05
about because it does add complexity. But I would
23:08
say that, you know, a 5 ,000 -person clinical
23:11
trial with, you know, a couple of hundred sites
23:14
is by its nature complex. And a complex structure,
23:19
I think, is best suited to deal with it. On the
23:21
other hand, you know, a phase one study or a
23:24
three -site phase two study just doesn't need
23:27
all this stuff. So taking into account the big
23:30
picture, I think it would be helpful to put forward
23:32
some ideas and move the discussion along. Well,
23:39
that would be terrific. Phase three trials all
23:41
in, I assume. Bigger trials, I think, is what
23:45
Rob is saying. Multi -site, multiple sites, multiple
23:49
stakeholders in this. But terrific discussion,
23:55
Scott. You stimulated this conversation. I think
23:58
we've got some work to do in the Collaboratory
24:01
here, but I want to thank my colleagues here
24:04
today with coming on to the Heart of the Matter
24:07
podcast of the HFC. And let's see if we can push
24:11
the needle forward and get... get trials in a
24:14
better position, particularly the larger trials
24:16
going forward. Thank you very much. Thanks, Chris.
24:19
Thanks, Chris. Thanks, Rob. Thank you, Chris.
24:22
Thanks, everyone.