Latest / Cardio Pod: Heart of the Matter / Representation vs. Diversity in Clinical Trials

Transcript

Representation vs. Diversity in Clinical Trials · Cardio Pod: Heart of the Matter

• 0:06
  Good afternoon. I'm Chris O 'Connor. I'm principal
• 0:08
  investigator of the Heart Failure Collaboratory,
• 0:11
  and we're here with Heart of the Matter, Heart
• 0:13
  Failure Collaboratory podcast to discuss the
• 0:16
  very important and dynamic area in clinical trials
• 0:21
  of representation and diversity. And I'm here
• 0:25
  with a distinguished panel to... comment on this
• 0:29
  over the next 30 minutes to see if we can provide
• 0:31
  insight, a deeper understanding, and help guide
• 0:35
  clinical trials so that they can be of the highest
• 0:38
  value going forward. And so let me go around
• 0:40
  the room and ask you to introduce yourself, Dr.
• 0:43
  Batchelor. Yes. Hi, I'm Wayne Batchelor. I'm
• 0:51
  the president of the medicine service line at
• 0:53
  the Inova Health System. I happen to also be
• 0:56
  an interventional cardiologist and a clinical
• 0:59
  trialist really interested in this particular
• 1:01
  topic. So really pleased to be here, Chris. Thanks
• 1:03
  for inviting me. Thank you, Wayne. Dr. Christian
• 1:06
  Gilliard. Hey, I'm Christian Gilliard. I'm a
• 1:11
  resident at University of Miami Jackson Memorial
• 1:13
  Hospital. And I've been working with the Heart
• 1:16
  Failure Collaboratory for a year in the representative
• 1:18
  population subgroup. And it's been a real pleasure
• 1:22
  and it's a pleasure to be here. Thank you. Dr.
• 1:27
  Orly Vardeni? Hi, I'm Orly Vardeni. I'm a professor
• 1:31
  of medicine at the University of Minnesota. And
• 1:34
  I've been working with Dr. Fusat and Dr. O 'Connor
• 1:38
  in the Heart Failure Collaboratory and co -chair
• 1:41
  the representative populations. subgroup, along
• 1:45
  with Vanessa Bloomer from Inova as well. Pleased
• 1:50
  to be here. Thank you. Dr. Mona Fouzad. Hi, everybody.
• 1:54
  I'm Mona Fouzad. I'm an associate professor of
• 1:56
  medicine at Duke University. And I'm also part
• 2:00
  of the working group that has been referred to
• 2:04
  the representative populations working group.
• 2:09
  Thank you. Thank you for being here. The collaboratories.
• 2:13
  mission is to improve the value of clinical trial
• 2:17
  conduct so that we can get the highest quality,
• 2:22
  economically reasonable costs, but help the development
• 2:30
  process, the ecosystem of development of drugs,
• 2:35
  devices, biologics in a more favorable fashion
• 2:37
  with all the stakeholders around the table contributing
• 2:41
  to how this should be changed, how we can do
• 2:44
  better. And one of the areas that we've been
• 2:47
  tackling for years is representativeness and
• 2:50
  representation in clinical trials. And I'm going
• 2:53
  to start with Dr. Batchelor to help us because
• 2:56
  we're in a dynamic world right now where we're
• 2:59
  talking about diversity and representativeness.
• 3:04
  And the terms matter, the language matters, and
• 3:09
  yet we're trying to get to an endpoint that is
• 3:14
  better than we have been in the past and more
• 3:16
  meaningful as we go forward. Wayne, help us through
• 3:21
  the nomenclature and what we're trying to achieve
• 3:24
  in clinical trials. Well, thanks, Chris. And
• 3:28
  thanks for putting on this podcast, because I
• 3:30
  think this is exceedingly important. There have
• 3:33
  been a lot of changes over the last few years
• 3:35
  on the nature of this conversation. And I think
• 3:38
  definitions really matter at the outset of embarking
• 3:43
  on this kind of conversation. So we have diversity
• 3:45
  and we have representation. At the end of the
• 3:48
  day, I firmly believe... that we should be really
• 3:53
  thinking about the regulatory science and what
• 3:55
  matters to patients and to the data. Diversity
• 3:59
  is just simply the definition of diversity when
• 4:02
  you think of it as just establishing different
• 4:04
  participant groups in the clinical trial. So
• 4:06
  having some different racial groups, ethnic groups,
• 4:10
  obviously sex, age groups, and perhaps rurality
• 4:14
  versus urban groups represented. But that's not
• 4:17
  the same as representation. I think it's extremely
• 4:20
  important that we understand that representation
• 4:22
  is what we are trying to achieve in clinical
• 4:25
  trials, not purely diversity. Representation
• 4:28
  speaks to the composition of trial participants
• 4:31
  and how closely those representative patient
• 4:35
  cohorts that are enrolled in clinical trials
• 4:38
  align and reflect the intended use population.
• 4:42
  So in other words, if I'm doing a trial in hypertension,
• 4:46
  I might have a diverse group. that is represented
• 4:49
  maybe by 5 % to 7 % African -Americans, 5 % to
• 4:54
  7 % Hispanic patients. But that's not necessarily
• 4:57
  the same as representation because well over
• 5:00
  30 % of our population of hypertension patients
• 5:03
  in the United States self -identify as being
• 5:06
  Black. It has a disproportionate prevalence.
• 5:09
  So many have referred to this participant to
• 5:14
  prevalence ratio. where you look at any particular
• 5:18
  demographic group and you look at the clinical
• 5:20
  trial, how well that group is represented in
• 5:23
  the clinical trial relative to what's the prevalence
• 5:27
  or the representation of that particular demographic
• 5:31
  group in that particular disease state. So I
• 5:34
  think it's really important to understand the
• 5:36
  difference between diversity and representation.
• 5:38
  And what we should probably be speaking about
• 5:40
  is representation more, which really is how well
• 5:45
  does a clinical trial reflect the intended use
• 5:48
  population? Because that's where we're getting
• 5:51
  all of our information on safety and efficacy
• 5:54
  for a drug or device. And that should be really
• 5:56
  underpinned by adequate representation in the
• 5:59
  representative patient cohort included in the
• 6:01
  trial. And Wayne, that is so well said. And I
• 6:06
  think that's really where we want to get. And
• 6:09
  if you think about the history of where we've
• 6:12
  been in clinical trials, we've often had these
• 6:16
  mufflers or these restrictions on inclusion,
• 6:19
  exclusion criteria. And so that you and I know
• 6:22
  when we did trials in myocardial infarction or
• 6:26
  heart failure, we would end up at the end of
• 6:28
  the day with an average age that was well under
• 6:32
  what we would anticipate the intended population
• 6:37
  or the population at hand that is fully at risk
• 6:40
  for that disease state. And so one area that
• 6:44
  we, I think, need improvement is in the area
• 6:48
  of lifespan representation and particularly taking
• 6:54
  down the barriers of saying, well, you can't
• 6:57
  get into clinical trial if you're over the age
• 6:59
  of 75. We've done better. I think one of the
• 7:04
  areas that I was impressed to see, Wayne, in
• 7:07
  the trials you've been involved with are the
• 7:09
  TAVR trials. The average age of the TAVR trials
• 7:12
  really do represent the intended population.
• 7:15
  Tell us a little bit about that. Well, it's a
• 7:18
  great comment. For a while, in the first series
• 7:22
  of patients enrolled in certainly the early trials
• 7:25
  of TAVR, they were over age 80. This was by design
• 7:29
  because, as you know, the first indication for
• 7:31
  transcatheter aortic valve replacement was in
• 7:34
  those patients who were deemed to be at prohibitive
• 7:37
  or high risk. of surgery. So age was a predominant
• 7:40
  feature in those patients. Still today, though,
• 7:43
  if you look at the average age of patients undergoing
• 7:45
  these commercially available procedures, they're
• 7:48
  still about age 80. So we're still seeing the
• 7:52
  devices applied quite commonly in older adults.
• 7:56
  And that also speaks to some of the guidelines
• 7:58
  that use age. as ways of distinguishing between
• 8:02
  different levels of evidence for supporting either
• 8:06
  surgery or transcatheter aortic valve replacement
• 8:08
  in our guidelines. So there are some things that
• 8:11
  are baked into this that relate to how we define
• 8:14
  the appropriate use of transcatheter aortic valve
• 8:17
  replacement and how it actually came into play.
• 8:20
  It really came into commercial use because of
• 8:22
  the need to come up with a solution in our older
• 8:25
  patient population. Some of that's by design
• 8:28
  and by the need to actually offer a unique solution
• 8:33
  that's lower risk and less invasive for our older
• 8:35
  patients. But having said that, I'm a firm believer
• 8:39
  that we've got to have age representation in
• 8:42
  our trials more uniformly. You know, the way
• 8:46
  that we metabolize drugs changes with age. Our
• 8:49
  likelihood of having a risk of a bleed or a vascular
• 8:53
  complication is remarkably dependent on age and,
• 8:56
  by the way, also on sex. We recognized in the
• 9:00
  early TAVR years that women had about a double
• 9:02
  risk of vascular complications and were more
• 9:06
  likely to have a serious bleed or vascular complications
• 9:08
  from these procedures. So if we don't have adequate
• 9:11
  representation of older adults and women in some
• 9:14
  of these transcatheter therapies, we're grossly
• 9:17
  underrepresenting or underappreciating some of
• 9:20
  the safety signals and certainly efficacy as
• 9:23
  well in these trials. We'll miss the mark. And
• 9:28
  getting to the point of representation, Orly
• 9:32
  and Christian, you've been working on a project
• 9:35
  looking at publicly available NIH trials to see
• 9:40
  if we can develop a representation index, an
• 9:45
  indice, and Dr. Blumer's been involved with this
• 9:48
  and Dr. Fouzant, taking the population. of that
• 9:53
  disease state and then seeing how did the different
• 9:56
  sites and geographies around the United States
• 10:00
  perform in recruiting patients that reflect the
• 10:05
  population at risk. Tell us a little bit about
• 10:09
  that, Orly. I'm actually going to let Christian
• 10:15
  tell us about it because he's been intimately
• 10:18
  involved and knows the data much better and can
• 10:21
  speak to it. better than I could. So please go
• 10:25
  ahead. Christian, start us off. Yeah, definitely.
• 10:29
  So, well, I've had actually the privilege of
• 10:32
  training outside of the U .S. also. And I think
• 10:35
  this is an important discussion because when
• 10:38
  I used to train outside, the question about clinical
• 10:42
  trials, when we were discussing results, was
• 10:45
  often, well... do these results matter if it
• 10:49
  doesn't adequately represent the population that
• 10:51
  we're treating here? And it took me by surprise
• 10:54
  when I came to the U .S. that we were also asking
• 10:58
  the question, and I'm training in Miami, but
• 11:00
  we were also asking the question of, did these
• 11:04
  trials adequately represent the population that
• 11:07
  we're treating in Miami? And so that's sort of
• 11:09
  the question that we had at the Hartford Recruit
• 11:12
  Aperture. When we were trying to initially study
• 11:15
  if there were regional differences in enrollment
• 11:19
  of patients for heart failure clinical trials.
• 11:22
  And we've seen in abstracts that we've published
• 11:25
  that those differences do exist. And of course,
• 11:29
  the question after that was, well, there might
• 11:32
  be regional variations just because the population
• 11:35
  of those trial sites are also... diverse and
• 11:39
  different from like between regions. And so then
• 11:43
  we came up with a question of how can we standardize
• 11:49
  or how can we find a way through participation
• 11:52
  to prevalence ratios to standardize a way to
• 11:57
  evaluate trials and trial sites to see if the
• 12:01
  representation was adequate, as we said, compared
• 12:06
  to the population that we were training. Well,
• 12:11
  and this is such an important point, Orly and
• 12:14
  Wayne, I want you to comment on this because
• 12:16
  we have this archaic approach to clinical trials.
• 12:20
  We identify sites basically on historical performance.
• 12:27
  Maybe they performed well in a different disease
• 12:30
  state. And so we gather these sites and then
• 12:34
  the sites are recruiting patients, you know,
• 12:38
  through scanning of charts or through subspecialty
• 12:42
  clinics, heart failure clinics and things like
• 12:44
  that. So there's enormous biases in who's getting
• 12:48
  into the trial. People who live in rural communities,
• 12:52
  people who live without transportation, people
• 12:55
  who don't have access to the subspecialty clinics
• 12:58
  are not being identified. And so already we have
• 13:02
  these types of barriers, as we were talking earlier,
• 13:04
  how artificial intelligence may help us take
• 13:07
  away some of those biases in identifying patients.
• 13:10
  But Wayne, you've published on this. Orly, you've
• 13:13
  been thinking about this a lot. How do we get
• 13:16
  better representation in our clinical trials
• 13:19
  at the site level? Might ask Orly to speak first
• 13:25
  and I'll chime in afterwards. Sure. I think one
• 13:28
  of the... points that you raised is crucial,
• 13:31
  which is we're always seeing the same types of
• 13:35
  or the same even sites in various clinical trials.
• 13:39
  And those sites tend to be large academic centers,
• 13:43
  which may not reflect the population at large
• 13:48
  and certainly not a rural population. So it's
• 13:51
  really important to try to engage sites at. different
• 13:57
  locations that are not necessarily tied to an
• 14:01
  academic institution, but are potentially standalone
• 14:04
  sites that serve a diverse or representative
• 14:09
  population around them to be able to draw from.
• 14:14
  And that will hopefully increase the representativeness
• 14:19
  overall. But I think one of the key ways to do
• 14:24
  this is to expand the the types of sites that
• 14:28
  we accept into clinical trials as performance
• 14:31
  sites. Yeah, so important, that topic. And it's
• 14:39
  one of the reasons, you know, that Wayne and
• 14:41
  I came to Inova because we said this is, you
• 14:44
  know, outside the traditional academic walls,
• 14:47
  but there's enormous population that has been.
• 14:51
  underserved by not having the clinical trial
• 14:54
  infrastructure in place in cardiology. Wayne?
• 14:58
  Yeah, Chris, you know, this is such a critical
• 15:01
  topic and a really important point. I'm so excited
• 15:05
  to see Christian, Orlo, Mona, the Heart Valve
• 15:08
  Collaboratory, and others working on this because
• 15:11
  I think we can do a lot better. I think we've
• 15:13
  been sort of trying the same experiment over
• 15:15
  the last 20 years. And not really getting, not
• 15:19
  really moving the needle. And what I mean by
• 15:21
  that is instead of having a really sort of scientific
• 15:25
  approach to solving this problem, we've sort
• 15:28
  of relied on a lot of sort of typical speech.
• 15:32
  You know, we've got to enroll more. Let's talk
• 15:35
  to these patients. Well, if you're going to sites
• 15:37
  that don't have access to those patients, you're
• 15:39
  just not going to get those patients enrolled.
• 15:41
  And I think what Christian and Orly have stated
• 15:45
  is quite relevant in that regard. There are about
• 15:48
  17 different things we can do to improve clinical
• 15:52
  trial participation in terms of enhancing underrepresented
• 15:56
  groups. The National Academy of Science has a
• 15:58
  very nice paper on this. But one of the levers
• 16:01
  that we have not pulled is exactly what Christian
• 16:04
  had referred to, which is taking advantage of
• 16:07
  the natural heterogeneity of our sites and understanding
• 16:11
  where to choose sites to participate in clinical
• 16:16
  trials. If we're doing a study that has to have
• 16:19
  a large number of Hispanic participants or African
• 16:26
  -American patients or other groups. We have to
• 16:31
  go where those patients actually live and to
• 16:34
  institutions that treat them. And some of these
• 16:36
  may be urban. Some of them may be clustered more
• 16:39
  in the southeast, especially for black populations
• 16:41
  in the United States. And we can't exclude the
• 16:44
  western part of the country, especially the southwest,
• 16:47
  to ensure that we have access to the Hispanic
• 16:50
  population. And I love the way that Christian
• 16:53
  is and you all are all starting to talk about
• 16:56
  this, which is. applying quantitative methods
• 16:59
  to it that fit nicely within our umbrella of
• 17:02
  regulatory science, participant to prevalence
• 17:06
  ratios, and understanding how the site selection
• 17:09
  and other levers that we can pull can move that
• 17:13
  PPR in the right direction so that we're getting
• 17:16
  good representation, PPRs of over 0 .8, for example.
• 17:20
  for all of our key subgroups. I think we can
• 17:22
  do a lot better. And I'm quite optimistic that
• 17:25
  you're going to see some changes over the next
• 17:26
  five years that will move the needle significantly
• 17:28
  in this realm. I think we're finally figuring
• 17:30
  it out a little bit better than we did. Well,
• 17:33
  well stated, Wayne. And I think that's really
• 17:36
  going to be one of the emphasis is to use science,
• 17:38
  quantitative information to move this field forward.
• 17:42
  Dr. Fuza, what are your last thoughts on this
• 17:45
  topic as we wrap this up? Yeah, I absolutely
• 17:50
  agree with everything that's been said and so
• 17:53
  nicely stated. I think I love the concept of
• 17:58
  the PPR, participant to prevalence ratio, you
• 18:03
  know, understanding where those patients are.
• 18:05
  And then also maybe looking a little deeper at,
• 18:09
  you know, is there a biologic reason that we
• 18:12
  need to have these, you know, investigations?
• 18:16
  Is there a genetic? difference in the people
• 18:19
  being treated or the way they handle a drug or
• 18:22
  device, the way Wayne outlined in the beginning,
• 18:27
  you know, are there bigger risks for certain
• 18:29
  groups? And so I think really bringing the science,
• 18:34
  emphasizing the scientific piece and understanding
• 18:37
  the biology is a really important aspect here
• 18:41
  as well. Well. Thank you very much, my esteemed
• 18:47
  colleagues and panelists. As always with these
• 18:50
  podcasts of the Collaboratory, we identify a
• 18:54
  lot of issues. We don't always solve all the
• 18:56
  answers, but we're moving the needle forward.
• 18:59
  So I'm particularly pleased with the Heart of
• 19:02
  the Matter, Heart Failure Collaboratory effort
• 19:04
  in this space to how can we do better. and making
• 19:10
  sure our clinical trials are representative of
• 19:14
  the patient population that we are studying,
• 19:18
  that we're looking at these different therapies.
• 19:20
  And you've brought some significant light to
• 19:24
  this important problem. So I want to thank you
• 19:27
  and thank our audience for being here this afternoon.
• 19:30
  Have a great day.