Latest / The BioHub - by Avetix / Episode 223 - Brent Richards -Founder & CEO | 5 Prime Sciences

Transcript

Episode 223 - Brent Richards -Founder & CEO | 5 Prime Sciences · The BioHub - by Avetix

• 0:06
  Welcome everyone to another episode of the Biohub podcast.
• 0:09
  Today I'm joined by Doctor Brent Richards, endocrinologist,
• 0:13
  Professor of Medicine at McGill University, and founder and CEO
• 0:17
  of Five Prime Sciences. Brent has spent his career
• 0:20
  applying to Human Genetics and large scale omics data to
• 0:24
  improve one of drug discovery's biggest challenges, identifying
• 0:28
  the right biological targets before programs enter the
• 0:31
  clinic. Through 5 Prime Sciences, he
• 0:33
  works with pharmaceutical and biotech companies to combine
• 0:36
  causal human biology with AI enabled approaches to improve
• 0:40
  confidence in target selection and ultimately increased chances
• 0:45
  of clinical success. Brent, welcome to the podcast.
• 0:49
  It's really kind of you to have me on the podcast, Amber.
• 0:52
  Amazing. In today's episode, we're going
• 0:55
  to be discussing why so many drug development programmes
• 0:58
  fail, the growing role of Human Genetics in drug discovery and
• 1:02
  where AI can have the greatest impact.
• 1:05
  But Brent, to start us off, can you tell us a little bit more
• 1:07
  about that, your background and what led you to founding 5 Prime
• 1:11
  Sciences? Yeah, sure, Amber.
• 1:14
  So I started my training in Human Genetics as an undergrad
• 1:19
  at McGill University in Montreal.
• 1:21
  And I distinctly remember being so excited and curious learning
• 1:26
  from one of my mentors, A pediatrician named Charlie
• 1:29
  Scriber, that I, I never really had to study in his course
• 1:32
  because I just kept thinking about the material.
• 1:34
  And this was about 6 years before the first draft of the
• 1:37
  Human Genome Project was completed.
• 1:39
  And, and I, I guess I was like a lot of people at that time had
• 1:42
  dreamt that it might be possible to understand the genetic causes
• 1:46
  of diseases and particularly common diseases in her lifetime.
• 1:50
  So after I got through medical school and Graduate School
• 1:52
  training, I returned to Amygdala's professor and I came
• 1:56
  back to that very simple question and had the opportunity
• 1:59
  to work on some of the earlier and earliest genome wide
• 2:03
  association studies. And I never really actually
• 2:06
  thought much about starting a company, but I became
• 2:09
  increasingly amazed that over time we could actually start to
• 2:11
  understand why some of our patients were getting sick.
• 2:15
  We could start thinking about intervening upon disease
• 2:17
  pathways that had recently been illuminated by Human Genetics.
• 2:21
  And maybe in some cases we could actually start helping patients
• 2:25
  where previously we could just offer them a diagnosis.
• 2:28
  So working as a professor and endocrinologist, I had the
• 2:32
  exciting opportunity to grow an academic lab.
• 2:34
  We were able to contribute to scientific literature, but it
• 2:38
  became apparent that it was difficult to make long term
• 2:41
  investments in academia and actually bring this through the
• 2:45
  darkest miles, which is through drug discovery to help patients.
• 2:49
  So our friends and colleagues who now had worked moved to
• 2:53
  industry kept asking us in our lab task to help them.
• 2:57
  We had the opportunity to watch GSK and help them make a large
• 3:01
  scale investment in Human Genetics and help them develop
• 3:04
  the capacity to understand human disease causes and sometimes
• 3:08
  treat them. And so with that as a pull in
• 3:12
  industry, we spun 5 prime out of our academic lab really with the
• 3:15
  mission to help other companies accelerate their drug
• 3:18
  development programs using these tools.
• 3:21
  Yeah. No, that's a really great
• 3:22
  background. And what would you say the kind
• 3:24
  of biggest challenges that you're trying to, what was the
• 3:28
  biggest challenge in drug discovery that you're aiming to
• 3:31
  solve? Yeah.
• 3:33
  So I when we, when we think about solving problems, we
• 3:37
  really try to anchor our work and what is the mission that
• 3:41
  we're trying to accomplish? Because if we can define the
• 3:43
  mission, then we can pick which problems to try to solve.
• 3:46
  So the mission's really pretty simple.
• 3:49
  And it's been exactly the same as my personal mission since I
• 3:53
  started my career. It's simply to use Human
• 3:56
  Genetics to improve the care of my patients.
• 3:58
  So so many of the disease remain untreated and most have fairly
• 4:03
  paltry treatment options. And genetics and increasingly AI
• 4:07
  informed genetics will have the opportunity to address this
• 4:10
  challenge. So 5 Prime tries to exist at
• 4:14
  that decision layer between human genetic data and drug
• 4:17
  development. And we feel that our mission is
• 4:20
  to better inform these decisions so as to enable trials to be
• 4:24
  successful and help our patients get better treatment options.
• 4:27
  Well, there's been an amazing investment in AI in the
• 4:31
  industry, and this is predominantly focused on
• 4:34
  molecule design, protein engineering, and refining
• 4:37
  medicines. This is really the first foray
• 4:41
  of AI into the field, and I think that it's been made for
• 4:44
  some psychological and social reasons that aren't always
• 4:47
  scientific. So patents can be filed on
• 4:50
  medicines, and if it's patentable, then it's often
• 4:52
  fundable, even if that's actually not often.
• 4:56
  The cause of drug development failure is getting the medicine
• 4:59
  right. The most common cause of drug
• 5:01
  development failure is just getting the target wrong.
• 5:04
  So I think that as the industry pivots to answering some of the
• 5:11
  more fundamental problems of drug development, which is what
• 5:15
  are the best targets that we should be treating, that we can
• 5:19
  start to be able to focus some of the arsenal that AI has
• 5:23
  developed upon these problems. So we're trying, we're applying
• 5:27
  AI to solve this more causal human biology problem.
• 5:32
  It's been an incredible tool that we're using very
• 5:35
  aggressively at 5 Prime to drive home our mission.
• 5:37
  We've helped develop novel algorithms and methods that
• 5:41
  allow us to partly solve central problems in drug development,
• 5:44
  such as which targets influence disease progression and not just
• 5:48
  susceptibility. Because our medicines actually
• 5:50
  treat progression of disease, they're not often preventing the
• 5:55
  onset of disease and can we actually predict drug
• 5:58
  development success for a specific target for a specific
• 6:01
  disease and it is working so we are able to be able to shed
• 6:06
  light into some of those field wide problems.
• 6:08
  Yeah. And obviously despite all of
• 6:11
  those advances as well, the majority of development programs
• 6:14
  still fail. Why do you think that is?
• 6:16
  Is that on the target ID side as well?
• 6:20
  Well, look, I mean, there's obviously lots of reasons why
• 6:23
  there is failure and the failure is truly remarkable.
• 6:27
  We've done a lot of work trying to understand drug development
• 6:30
  failure at 5 prime. We've worked on it for about the
• 6:33
  past five years and I think we have a few clear answers.
• 6:37
  One is something that has been described in the literature and
• 6:41
  not everyone is aware of. Is that the most, by far the
• 6:45
  most common reason why drug development programs fail?
• 6:47
  It is it is it phase two and it's just because of lack of
• 6:51
  efficacy. It's not that the drug didn't
• 6:53
  engage with the target. It's not most often that the
• 6:56
  drug had horrible side effects. It's not that the business case
• 6:59
  has changed. It's simply we gave the drug and
• 7:02
  nothing happened. That's quite remarkable given
• 7:05
  the billions of dollars spent per year on drug development.
• 7:08
  So we took a deep look at this problem.
• 7:11
  We realized when we were thinking this through that we're
• 7:14
  actually trying to solve 2 problems.
• 7:16
  The 1st is that there's a set of decisions and predictors that
• 7:20
  influence whether or not a drug development team actually
• 7:22
  launches the clinical trial. And then there's a totally
• 7:26
  separate set of predictors that influence whether or not that
• 7:28
  trial is going to be successful. So we were, I guess, somewhat
• 7:33
  surprised to realize that the single most important predictor
• 7:36
  of actually launching a trial is simply the number of
• 7:40
  publications based upon a target.
• 7:43
  Simply put, people do trials and targets that they know or that
• 7:47
  they've worked on in the past, but the number of publications
• 7:51
  that have been generated on the target has absolutely no effect
• 7:55
  on whether or not the trial would be successful.
• 7:58
  So studying what we know is a great way to get funding to do a
• 8:02
  trial from investors or academia, but it has no impact
• 8:07
  on whether or not that trial would be successful.
• 8:09
  That's the first problem. The second problem, once we've
• 8:13
  solved that psychological and social problem, is actually
• 8:16
  trying to improve that efficacy failure.
• 8:18
  And the single best predictor of a successful trial is whether or
• 8:22
  not there's human genetic evidence that the target
• 8:24
  actually causes disease in humans.
• 8:27
  But we've known this as a field field for several years.
• 8:32
  But not all human genetic evidence is the same.
• 8:35
  We found tremendous differences in value across the different
• 8:38
  types of Human Genetics evidence and their impact on drug
• 8:41
  development success. And then we use that information
• 8:44
  to help our partners get the best sight on whether or not a
• 8:47
  program will be successful. So the question I think isn't
• 8:51
  simply in order to reduce failure, isn't simply is this a
• 8:55
  genetically validated target, it's what are the types of human
• 8:59
  genetic evidence that support this drug development program?
• 9:03
  Yeah, Human Genetics, then, has become an increasingly important
• 9:07
  part of target discovery. How is it changing the way
• 9:10
  companies identify and validate new therapeutic targets?
• 9:15
  Yeah. So I think genetics used to be a
• 9:18
  late stage checkbox. So at investment boards, pharma
• 9:22
  companies would ask does this target have genetic support be a
• 9:26
  yes or no answer. And that was often after most of
• 9:29
  the discovery work was done. So people have developed the
• 9:31
  medicine, they try to understand the mechanism, but now we're
• 9:34
  seeing companies push this decision much earlier into the
• 9:36
  pipeline. So investors at the most
• 9:39
  sophisticated venture capital groups are asking this prior to
• 9:41
  writing a check. People are becoming increasingly
• 9:44
  aware that that there's not one answer to that question because
• 9:47
  there's so many different types of human genetic evidence.
• 9:49
  Some of them have so much more to add to that question than
• 9:53
  others. And this has a great impact on
• 9:56
  the variability in probability of drug development success.
• 10:00
  So translating this information into decision grade intelligence
• 10:06
  is what has helped 5 Prime make an impact with our partners and.
• 10:11
  That's some really interesting insights there across the
• 10:14
  market. And so 5 Prime combines Human
• 10:17
  Genetics with other omics technologies to support it's
• 10:21
  partners. How do these different data
• 10:24
  sources work together to build confidence in a potential drug
• 10:28
  target? Yeah.
• 10:31
  So a genetically validated drug target does not a good medicine
• 10:35
  make. Genetics helps you answer causal
• 10:38
  biology questions such as what's the whole body effect of
• 10:42
  perturbing this target? But we need to add on multiple
• 10:45
  levels of evidence on top of Human Genetics to be able to
• 10:47
  answer other important questions such as do I modulate the
• 10:51
  circulating level of the target? Which direction do I modulate
• 10:55
  that target to get the desired therapeutic effect?
• 10:58
  What tissue should I point my medicine arsenal at?
• 11:01
  Is there a specific cell type within that tissue that I can
• 11:04
  that I can attribute to the causal effect of the drug
• 11:07
  target? So layers such as proteomics,
• 11:10
  single cell and spatial are increasingly helping us to be
• 11:14
  able to answer these questions for our partners.
• 11:16
  Yeah. And looking ahead then, what
• 11:19
  scientific or technology advances do you think will have
• 11:22
  the biggest impact on improving that drug discovery over the
• 11:26
  next 5 to 10 years? Yeah, I, I mean, I think
• 11:30
  there'll be lots. First ancestral diversity is
• 11:34
  really exciting. So just think about it.
• 11:36
  So Homo savings have existed on the planet for about 300,000
• 11:39
  years. The vast majority of human
• 11:42
  genetic, that data that we have now is from people of European
• 11:45
  descent. That population, depending on
• 11:48
  how you define it, is only about 50,000 years old.
• 11:51
  So as people of different ancestries participate in
• 11:55
  studies, we'll get a much better understanding of the entire
• 11:58
  gamut of human genetic variation.
• 12:00
  That will be really helpful. But I think AI applied to causal
• 12:03
  inference, not just pattern finding, separating moves with
• 12:07
  disease from causes disease at a much greater scale, much
• 12:11
  different types of tissues and cells will make a better
• 12:14
  evidence based decisions for pharma.
• 12:17
  So I mean, I think reflecting there's not going to be a single
• 12:21
  shift. It's not just one technology,
• 12:23
  but it's whether companies can build the standing capability
• 12:27
  across the whole pipeline versus buying one slice and applying it
• 12:32
  one at a time, but being able to apply the whole tool kit to
• 12:36
  improve the probability of drug development success.
• 12:39
  Yeah, no, amazing. Well, thanks so much for giving
• 12:41
  me that overview as well. And you know, just to round us
• 12:44
  off and a question that we ask all the guests on the Biohub
• 12:47
  podcast, if you could have any have, if you could have dinner
• 12:51
  with any scientist or innovator, past or present, who would it be
• 12:55
  and why? Yeah, So I think that's easy.
• 12:57
  It's it's Archibald Garrett. So maybe not known to a lot of
• 13:01
  people in the field, but he was a physician in the early 1900s
• 13:05
  who cared for patients at Saint Barts in London.
• 13:08
  He treated rare diseases and most physicians ignored these
• 13:12
  diseases, medical curiosities and and he really looked at them
• 13:16
  as clues to help us be able to understand what's causing
• 13:19
  disease. So his simple curiosity led him
• 13:23
  to believe that natural genetic variation could reveal how human
• 13:27
  biology works without really even knowing anything about
• 13:29
  genetics. It was a remarkably modern idea
• 13:33
  decades before anybody had even discovered DNA.
• 13:36
  So just by way of example, one of my favorite stories is that
• 13:42
  he encountered a family of patients who had a very weird
• 13:47
  medical curiosity. And that is is that if you let
• 13:51
  their urine stand overnight, it would turn black.
• 13:55
  And most physicians were felt, well, that's kind of neat, but
• 13:59
  who cares? And he just kept asking why.
• 14:03
  And that observation led him to the concept of inborn errors of
• 14:09
  metabolism and in many ways laid the intellectual foundation for
• 14:14
  precision medicine, being able to understand causes of disease
• 14:17
  in human using Human Genetics. And I now we would say that the
• 14:23
  patients have a mutation in a gene called HGD.
• 14:27
  But he was able to foresee all of this without having any
• 14:29
  insight as to what DNA is, what are the modes of inheritance,
• 14:33
  what even what genes are. So I think that it would be
• 14:37
  exciting for me to to meet him because at 5 prime, we're doing
• 14:40
  the same thing, but with a pretty humbling tool chest.
• 14:42
  I mean, we have data from millions of humans, not just a
• 14:45
  small family. We're not just looking at
• 14:47
  medical oddities. We're looking at very common
• 14:49
  diseases that are killing people.
• 14:51
  We have computational resources and methods which were frankly,
• 14:54
  totally inconceivable in the early 1900s.
• 14:57
  But we're essentially driven by the exact same curiosity to
• 15:00
  solve the exact same problem. We're trying to learn from
• 15:03
  nature's genetic experiments to help our patients by discovering
• 15:08
  new medicines. So I'd love to have dinner with
• 15:10
  him, tell him what's happened over the last approximately 100
• 15:13
  years and ask him what he would tackle next.
• 15:16
  Yeah. He sounds so impressive and I
• 15:18
  love that kind of thought on on the curiosity piece.
• 15:21
  Well, Brent, thank you so much for joining us on the bio
• 15:24
  podcast today. Excellent.
• 15:27
  Thank you so much, Amber. I really enjoyed the opportunity
• 15:30
  to talk to you.