AI transcript
0:00:03 Hi, and welcome to the A16C podcast.
0:00:04 I’m Hannah.
0:00:06 The federal agency known as the FDA,
0:00:07 or the Food and Drug Administration,
0:00:09 was born over 100 years ago
0:00:12 at the turn of the Industrial Revolution
0:00:14 in a time of enormous upheaval and change
0:00:16 and rapidly emerging technology.
0:00:19 All of those things could be said to be just as true today.
0:00:23 From CRISPR to synthetic biology to using AI and medicine,
0:00:24 our healthcare system is undergoing
0:00:27 massive amounts of innovation and change.
0:00:28 This wide-ranging conversation
0:00:31 between Principal Commissioner of the FDA,
0:00:32 Amy Abernathy, and Vijay Pandey,
0:00:34 general partner at A16Z,
0:00:38 took place at A16Z’s annual summit in 2019,
0:00:41 and covers everything from gene editing your dog
0:00:43 to tracking the next foodborne outbreak,
0:00:45 how advances in bioengineering
0:00:47 are transforming healthcare, clinical trials,
0:00:48 and drug development,
0:00:50 and how the federal agency is evolving
0:00:53 to keep pace with the scientific breakthroughs coming
0:00:55 while staying true to its core mission
0:00:58 of assessing safety and effectiveness for consumers
0:01:00 in the world of food and medicine.
0:01:02 – So thank you so much for joining us.
0:01:04 – Terrific to be here, hello.
0:01:06 – So, you know, in thinking about how I start this,
0:01:08 I was thinking about the origins of the FDA.
0:01:13 So the FDA started in 1906, 113 years ago.
0:01:17 And it’s interesting to think about that time
0:01:20 because, you know, it’s turn of the previous century,
0:01:22 a time of a lot of tumult, innovation,
0:01:26 technical change, an Industrial Revolution,
0:01:28 you know, things that actually really were
0:01:31 the driving forces to create the FDA.
0:01:33 And like, here we are, another turn of the century,
0:01:34 another Industrial Revolution,
0:01:36 another amount of tumultuous change.
0:01:39 You know, what are the needs of the FDA right now?
0:01:42 And, you know, is its core mission really still relevant?
0:01:44 – Is the FDA still relevant?
0:01:45 – Yeah.
0:01:48 – So, I’m gonna go back that 113 years
0:01:50 and the time the FDA was formed
0:01:53 and continues to be the largest consumer protection agency,
0:01:55 was formed out of a hundred laws.
0:01:57 I think that the issue that was going on at the time
0:02:00 was unhygienic conditions in the Chicago stock yard.
0:02:03 And you can imagine, there’s been a lot of responsibilities
0:02:06 of the FDA over time, phyllidimide, et cetera.
0:02:09 But practically speaking, the FDA is responsible
0:02:12 as a science-based agency to protect
0:02:15 and promote public health, including through making sure
0:02:18 that we have safe and effective medical products
0:02:20 to use every day with our patients,
0:02:22 as well as through promoting innovation.
0:02:24 – You know, your question was
0:02:26 whether or not the FDA is still relevant.
0:02:27 – Yes.
0:02:28 – And I would argue that in a time
0:02:31 of rapidly emerging biology,
0:02:34 when we’ve got more and more scientific innovations
0:02:37 and potential products coming to bear,
0:02:40 the need to make sure that we have an objective way
0:02:42 of assessing safety and effectiveness
0:02:45 and providing consumer confidence
0:02:48 that this treatment is appropriate for me,
0:02:50 it is a responsibility of the FDA
0:02:52 that actually has more responsibility, not less.
0:02:53 – Well, so, you know, in that context,
0:02:55 let’s talk about what the FDA looks like today.
0:02:57 In those older days, you know,
0:03:00 the data came to the FDA by the truckload.
0:03:02 You know, I’m just imagining like,
0:03:03 reams and reams and paper and so on.
0:03:05 And you know, and I’m just curious to get your take
0:03:08 from, you know, what does the current system look like?
0:03:10 And you know, could you take us through, you know,
0:03:11 how this works?
0:03:12 – So, you know, a couple of things.
0:03:13 I actually think that there was a time
0:03:16 when it probably came on horse and buggy, not just trucks.
0:03:17 – Yes, yes, yes.
0:03:19 – So, practically speaking,
0:03:21 I usually think about five key elements
0:03:24 in drug and biologic product development.
0:03:26 So, there’s the discovery phase,
0:03:28 then there’s time of preclinical development.
0:03:31 After you’ve done adequate preclinical development
0:03:33 in line with good laboratory practice,
0:03:36 then you’d submit an investigational new drug application
0:03:39 for a drug or a biologic to the FDA,
0:03:41 which gives permission then to start
0:03:44 clinical studies with people.
0:03:46 And a drug or biologic will go through
0:03:47 a series of clinical studies,
0:03:49 typically phase one through three,
0:03:51 although these days those lines are blurring
0:03:56 for exactly what drugs follow then by a new drug application
0:03:59 or biologics application of BLA to the FDA.
0:04:02 And then the final stage after FDA marketing approval
0:04:03 would be post-marketing assessment
0:04:05 and continuous surveillance
0:04:07 about this particular medical product.
0:04:09 So, that’s kind of the usual steps.
0:04:10 – Given that, you know,
0:04:13 what would you want to modernize about it?
0:04:16 You know, how do you take us into the next 100 years?
0:04:18 – Yeah, so, you know, I think that goes back
0:04:19 to the horse and buggy and the trucks
0:04:22 and how information has historically gotten to the FDA.
0:04:26 So, I came from the landscape
0:04:29 of a health tech startup company focused on data.
0:04:32 I was recruited to FDA with the expectation
0:04:35 that I was gonna show up and focus specifically
0:04:38 on digital innovation for the FDA
0:04:43 and sort of went smashing into a realization
0:04:45 that in fact, the way the majority of our applications
0:04:49 still come in is through PDFs
0:04:53 or sort of essentially large digital representations
0:04:56 of what used to come in on trucks.
0:04:59 We now receive most of our drug applications,
0:05:03 but not all in some kind of electronic format.
0:05:05 As a matter of fact, just two weeks ago,
0:05:09 I had to approve a book of work in orphan diseases
0:05:11 where we’re still getting things on paper as one example.
0:05:15 But most things come in as a digital application
0:05:18 also with some digital data that gets stored at FDA.
0:05:22 And as I think about where we’re going in the future,
0:05:26 practically speaking, in order to become more efficient FDA,
0:05:28 we’re gonna need to receive more and more
0:05:31 of those applications in digital formats
0:05:33 that start to represent structured data,
0:05:36 structured data that we can review at scale
0:05:38 and also structured data that allows us
0:05:41 to now continuously surveil medical products
0:05:42 in a better way in the future.
0:05:45 And so what I realized was that if I was going to be
0:05:47 a person focused on data at FDA,
0:05:50 I was gonna need to also think about how do we modernize
0:05:53 the FDA’s underlying infrastructure to take that forward.
0:05:55 And that’s why I took on the CIO job.
0:05:56 – Yeah, in addition to the infrastructure,
0:05:59 it’s interesting to think about the culture.
0:06:03 And because for example, if there’s a great drug
0:06:07 that nobody ever gets, nobody ever knows about it.
0:06:08 Let’s say there was a cure to cancer
0:06:10 but the FDA didn’t approve it.
0:06:13 There’s no outcry because no one ever knew about it.
0:06:16 But on the other hand, if the FDA lets something through
0:06:18 that actually has harmful effects,
0:06:20 thalidomide and other classic examples,
0:06:23 then there’s a huge backlash.
0:06:26 How do you build and how do you innovate in a culture
0:06:29 that has to deal with such strong asymmetries?
0:06:32 – So it’s interesting as you describe those asymmetries,
0:06:35 what you all heard describing is the practical reality
0:06:38 that that can make you very risk adverse, right?
0:06:39 – Yes, exactly.
0:06:41 I’m really worried that I’m gonna do something wrong
0:06:43 and I’m gonna flub up and that’s actually gonna have
0:06:47 very public impact and we see that all the time.
0:06:52 I think that in order to develop solutions
0:06:55 for regulatory innovation, then what you really have to do
0:06:57 is come up with flexible mechanisms
0:07:00 that also allow us to deal with the risks
0:07:03 but also take some risks when appropriate.
0:07:06 And so that means that as FDA,
0:07:08 one of the things that we focus on
0:07:11 is risk-based scientific decision-making
0:07:15 and trying to right-size the degree of review
0:07:18 and expectation with the potential risk
0:07:20 of this particular product.
0:07:22 And what do I mean by risk?
0:07:23 Sometimes there’s safety risks, right?
0:07:26 So the risk of, for example, hepatic failure
0:07:29 or the risk that the drug might take a person’s life.
0:07:32 Sometimes the risks of the size of the population impacted.
0:07:34 So you’re trying to balance the urgency
0:07:36 of this particular problem sitting in front of you
0:07:40 with the number of people where this product may impact.
0:07:45 Also, there’s risks of public perception and expectation
0:07:48 and then the last sort of set of risks is
0:07:49 where can you de-risk it?
0:07:50 – Exactly.
0:07:53 – So you can de-risk it by trying to make sure
0:07:55 that, for example, preconditions are met
0:07:57 as it relates to the manufacturing process.
0:07:59 You can de-risk it by understanding
0:08:01 in a consistent way toxicity.
0:08:04 You can de-risk it by having consistent expectations
0:08:06 around clinical effectiveness.
0:08:10 – So I think what’s interesting is to think about that
0:08:12 in even the other context of what the FDA does.
0:08:15 I think people often don’t realize that the FDA
0:08:17 isn’t just about, let’s say, approving drugs.
0:08:18 We think about clinical trials.
0:08:20 There’s a lot of things that you do
0:08:23 to protect American consumers.
0:08:25 And when we’re talking about it, it’s almost like,
0:08:28 I feel like you could have a show that’s like CSI FDA
0:08:31 or something like that where you have these investigations
0:08:32 of the crisis.
0:08:35 Like, you know, like sometimes it’s slow moving crises,
0:08:36 like the opioid epidemic.
0:08:39 How do you, like for a crisis like that
0:08:43 where it slowly sneaks up on us and then it’s too late?
0:08:45 You know, how does the FDA even think about that?
0:08:48 – Practically speaking, the FDA is responsible
0:08:49 for many types of medical products.
0:08:53 And any one of those can have a crisis, I’ve discovered.
0:08:55 So we have food and drugs.
0:08:57 We have biologics and devices.
0:08:59 We have animal food and drugs.
0:09:00 We have cosmetics.
0:09:03 We have nicotine based products, vapes.
0:09:07 And so the distribution of potential crises are real.
0:09:11 And as you think about something like the opioid crisis,
0:09:16 a problem that snuck up on us in many different ways.
0:09:21 And as I step back, and I came to the FDA in March,
0:09:23 so I’ve had sort of the opportunity to watch this
0:09:26 as an insider/outsider during this period of time.
0:09:29 You know, as information starts to accumulate,
0:09:32 that says we’ve got a really big problem here.
0:09:34 That information comes from the public.
0:09:37 It comes from across different places in government.
0:09:39 And now we need to step back as a nation,
0:09:41 but as also as an agency and say, what do we formally do?
0:09:45 And so as an agency, we put in place an action plan
0:09:48 that had several parts that focus
0:09:49 on what we’re responsible for.
0:09:52 And I’m gonna come back to that key point in just a second.
0:09:55 But then also asked, how does that interdigitate
0:09:57 with all the other plans that are going on
0:10:00 across government and also across the healthcare setting
0:10:02 to try and solve for?
0:10:05 So I’ll say the last piece that I’ve found very interesting
0:10:09 since I’ve been in government is that there are very clear
0:10:11 rules of the road of our authorities.
0:10:13 It took me a while to get used to that word,
0:10:15 but the word would be authorities.
0:10:16 – Authorities, yes, yes.
0:10:19 – Yeah, so our area is a responsibility
0:10:21 and sort of legitimate jurisdiction.
0:10:25 And practically speaking, with respect to the opioid crisis,
0:10:28 we need to go back and say, whereas FDA,
0:10:32 do we have authority to try and help resolve this problem?
0:10:36 So practically speaking, we can help reduce
0:10:39 the number of opioid tablets, for example,
0:10:42 a patient has access to after back surgery or knee surgery
0:10:45 in order to reduce the chance that this particular person
0:10:47 has access and becomes addicted in the first place.
0:10:51 As a second example, we can increase methods for access
0:10:55 for example, naloxone-based treatments in the field.
0:10:58 And we’ve done a number of projects to try and make sure
0:11:01 that there is patient-informed labeling
0:11:03 and other aspects in the field.
0:11:06 And then also, we can start to think about developing
0:11:08 and helping to develop new treatments for the treatment
0:11:10 of pain as well as new treatments for the treatment
0:11:13 of addiction and start to solve a problem on that side.
0:11:15 So as FDA, we have to stick in our swim lanes,
0:11:17 but then we have to think about how that grooves
0:11:19 with everything else across government
0:11:22 so that there’s more of a nationwide approach.
0:11:27 – Yeah, in that sense of handling the authority nature
0:11:30 of things, you’ve got to make some tough decisions,
0:11:32 like even things like contaminated food
0:11:35 going across the border, like you’ve got to inspect trucks.
0:11:37 How do you know which truck to look for?
0:11:40 – Yeah, so as I got to the agency,
0:11:42 I was surprised to find not only are responsible
0:11:47 for regulating about 20% of international GDP
0:11:51 as I mentioned, that sort of is cross a broad number
0:11:53 of products and the weed that we regulate
0:11:55 is different by product.
0:11:58 So about 15% of our food is imported.
0:12:00 We need to be able to make sure that the food
0:12:02 is appropriately safe.
0:12:05 It’s appropriately labeled that it’s legitimate
0:12:06 for sale in this country.
0:12:08 And so that means that if you’re sitting
0:12:13 at the border of Mexico, we need to basically investigate
0:12:15 trucks that are coming across the border
0:12:17 and look for violative products
0:12:19 that aren’t appropriate for sale in the United States,
0:12:22 either for safety concerns or commercial concerns.
0:12:24 How do you know which truck to look at?
0:12:26 And if we don’t get that right,
0:12:29 we can basically stop traffic for miles.
0:12:33 So we have in the case of inspecting trucks,
0:12:35 we have something called the Predict Program.
0:12:39 And the Predict Program is a 10 years old rules engine
0:12:42 that’s been written by some of the different centers
0:12:45 across FDA where the rules start to predict
0:12:50 which truck is most likely to have unsafe food.
0:12:51 Now you can imagine that if we wrote those rules
0:12:55 10 years ago, they might be old rules.
0:12:58 And it’s true that we update the rules every year,
0:13:00 but we do so by hand.
0:13:02 And as I think about trying to develop
0:13:03 a more modern agency,
0:13:06 can’t we update the way that we modernize those rules?
0:13:08 And so right now we’ve got an experiment going on
0:13:11 where we’re looking at machine learning based prediction
0:13:13 of which trucks we should inspect on the border.
0:13:18 And can we now use machine learning as a way to be smarter?
0:13:21 Importantly though, going back to your point
0:13:22 about we can’t get it wrong
0:13:25 because we lose consumer and confidence when we get it wrong.
0:13:28 We can’t just say machine learning is gonna be great,
0:13:28 let’s go.
0:13:31 We actually have to thoughtfully do the experiments
0:13:35 to say if we apply a new approach over the old rules engine,
0:13:38 are we going to now be able to improve
0:13:39 our inspection on the border?
0:13:41 – Yeah, no, I think it’s fascinating to imagine
0:13:43 that there is this world where they have to use
0:13:45 doing deep learning, machine learning to be able to do this.
0:13:48 You know, I think perhaps also we underestimate
0:13:52 maybe cases where maybe you have averted crises
0:13:53 that we never heard of.
0:13:55 You know, those might be the best TV shows.
0:13:56 I mean, are there any cases like that?
0:13:57 – As you were saying this,
0:13:59 I was trying to come up with some crises I could talk about.
0:14:01 That was actually what I was thinking.
0:14:04 So, you know, here’s one that I just recently learned about.
0:14:07 I think we’re all very worried about drug shortages.
0:14:08 I’m an oncologist by background.
0:14:10 I practiced in academia.
0:14:12 I took care of adults,
0:14:16 but certainly when we think about children with leukemia,
0:14:18 one of the drugs that’s currently in shortage
0:14:21 is a leukemia drug for pediatric leukemia.
0:14:26 And we try and think through how do we help
0:14:27 avert drug shortages?
0:14:30 And in 2018, we had something north of 50 drug shortages,
0:14:34 but the little known secret is that we helped to avert
0:14:36 over 160 drug shortages.
0:14:38 That’s because practically speaking,
0:14:40 we have developed a whole staff
0:14:42 focused on drug shortages.
0:14:45 We’ve tried to start to figure out ways to predict
0:14:47 what is going to cause a drug shortage
0:14:49 and try and intervene beforehand,
0:14:51 speaking directly to manufacturers.
0:14:54 It takes sometimes a while to build that muscle, right?
0:14:56 You can imagine, we have to first understand
0:14:57 what are the causes of drug shortages
0:14:59 and how are we gonna go after them?
0:15:02 But then practically speaking, once we do so,
0:15:04 we can help to avert that crisis.
0:15:06 I see the same kind of thing right now
0:15:07 in foodborne outbreaks,
0:15:10 where we have a call every morning at 9 a.m.
0:15:12 where we’re sort of talking about the things
0:15:13 that worry us for the day.
0:15:15 I’ve officially stopped eating.
0:15:17 – Wow, that’s not good news for any of us.
0:15:20 – ‘Cause I’m getting very concerned
0:15:21 about what foodborne outbreak there’s gonna be
0:15:22 through the day.
0:15:24 But there’s sort of like this continuous sensing
0:15:26 to try and avert problems before they come.
0:15:27 – Yeah, fantastic.
0:15:29 So let’s change the channel.
0:15:30 So we’re watching CSI.
0:15:31 Let’s switch to a different show.
0:15:33 Let’s watch, let’s get into maybe something
0:15:34 more like Star Trek.
0:15:37 So let’s talk about the future,
0:15:41 because the future that’s really becoming today,
0:15:45 like stuff that I remember like five, 10 years ago,
0:15:46 were things that I thought would be sci-fi,
0:15:51 like gene therapy, gene editing, CRISPR therapies.
0:15:55 It’s kind of crazy that I was actually talking
0:15:59 with my eldest daughter and she was finding
0:16:03 that you could get kits off of Amazon to do DIY CRISPR
0:16:06 and that she could make our dog glow in the dark.
0:16:09 She liked it not to make our dog glow in the dark,
0:16:12 but I think in the end I think it was just crazy
0:16:14 that this is the world that we’re living in.
0:16:17 And so, you know, there’s two sides of this to dive into.
0:16:20 So maybe the first side is like we’ll get to kids
0:16:21 and dogs glowing in the dark in a second,
0:16:25 but like for thinking about the clinical side of this,
0:16:28 you know, how does the FDA think about something
0:16:31 like CRISPR because it’s both gene editing,
0:16:35 it’s a therapy, there’s a delivery aspect.
0:16:38 You know, does this live with the FDA?
0:16:40 Does this live with the AMA?
0:16:42 You know, how do you even start to think about people
0:16:44 throwing these crazy new developments at you
0:16:48 that could radically transform medicine and cure disease,
0:16:50 but now has to really push the paradigm
0:16:52 for the FDA in new ways?
0:16:55 – So I think one of the things to go back to
0:16:59 is this issue of risk-based regulation.
0:17:02 You know, how are we gonna start to solve for this
0:17:05 in a way that appropriately has the right regulatory paradigm
0:17:10 for this problem at hand and aligns with the level of risk?
0:17:13 And as I think about this space, you know,
0:17:15 I think that we are living in a space
0:17:17 that gets closer and closer to customized
0:17:19 or individualized therapies,
0:17:21 the landscape of the end of one,
0:17:23 and how do we make sense of that?
0:17:26 And practically speaking, in order to get there,
0:17:29 you have to have some kind of framework
0:17:31 that you apply that says,
0:17:34 all right, before we talk about CRISPR specifically
0:17:39 or anti-sense-aligonucleotides,
0:17:41 you know, before we talked about any specific thing,
0:17:43 like what’s the framework when we’re going to apply
0:17:47 and how do we do so in a risk-based way?
0:17:50 And practically speaking, that includes,
0:17:51 what do we know about safety?
0:17:54 What do we know about safety in vivo and in vitro?
0:17:56 In animals and in the petri dish,
0:17:59 what do we know about biological plausibility?
0:18:01 You know, what’s our understanding
0:18:04 from a biology perspective that this would indeed work
0:18:06 in the way that we would expect it to work?
0:18:11 What do we have in terms of a predefined set
0:18:14 of expectations in terms of clinical outcomes
0:18:18 that we can monitor in a objective way
0:18:20 to understand whether or not this intervention
0:18:22 is making the difference that we expect it to make?
0:18:25 Also, what do we need to think about with respect
0:18:26 to whether or not this is going to apply
0:18:28 just to one individual person
0:18:32 or we might now start to apply and scale this approach
0:18:34 across multiple individuals?
0:18:36 And that’s actually going to start to balance
0:18:37 how much risk we’re going to take
0:18:38 for this particular scenario.
0:18:41 What can we think about consistency and manufacturing?
0:18:44 And manufacturing I think starts to become more and more
0:18:46 of an issue across this space.
0:18:48 And then practically speaking, what are the ethics?
0:18:50 Like what’s going to happen in the clinic?
0:18:53 We may not be responsible in our authorities for ethics,
0:18:55 but I think we’re responsible for at least consciously
0:18:57 thinking about what’s going to go on.
0:19:01 So as I think about this space of incredible new therapies
0:19:06 coming forward, we all need frameworks that we can apply
0:19:09 and where every one of us can look at through a different lens
0:19:11 and say, I can understand why that’s the order
0:19:12 and that’s how we’re going to start
0:19:13 to work our way through it.
0:19:15 – Well, so let’s drill down a little deeper
0:19:18 because the end of one framework sounds like
0:19:20 mind boggling from a therapeutic point of view,
0:19:23 but maybe there’s actually a presence.
0:19:26 So think about surgeries, like heart surgeries
0:19:28 are all kind of n equals one, people are different.
0:19:33 There’s some similarities and perhaps we’re starting to look
0:19:36 at things like CAR-T, not just as actually back up.
0:19:39 So CAR-T is in this sci-fi category.
0:19:42 You take T cells out of your body, out of your blood,
0:19:46 you re-engineer them to make them sort of supercharged
0:19:48 and you put them back into the patient.
0:19:50 And the results of that are just mind boggling
0:19:53 that tumors can melt away within days
0:19:55 and people are just literally cured of cancer.
0:19:59 And so CAR-T has sort of biopharma aspect.
0:20:02 Maybe it has like a sense of doing molecular surgery,
0:20:05 you know, like in these n equals one cases.
0:20:06 So I’m curious to get your sense,
0:20:09 like n equals one is not completely unprecedented,
0:20:12 but what are the things that we need to do
0:20:14 to bring it into the future?
0:20:17 – So, you know, not only is n one not precedent,
0:20:19 we kind of go back across medicine, across time,
0:20:21 we actually really started off as in a one medicine
0:20:24 and became more quantitative across time,
0:20:25 especially as we had interventions
0:20:27 that were applicable to populations,
0:20:28 including in the million.
0:20:30 So practically speaking,
0:20:33 we do have frameworks for in-of-one therapies.
0:20:34 You mentioned surgery.
0:20:35 Another one, you know, in the landscape
0:20:37 that I came from in cancer medicine
0:20:39 was bone marrow or sim self-transplant, right?
0:20:44 These are places where we needed to first have
0:20:46 the scientific innovation that goes along
0:20:48 with biological plausibility
0:20:51 to start to figure out how we’re going to move forward
0:20:54 with new techniques and treatments about what to do,
0:20:57 but then start to apply a systematized set of expectations
0:20:59 about how to refine this and get it right.
0:21:02 So if I go back to bone marrow transplant,
0:21:05 you know, we started to develop refined processes
0:21:09 to understand which patient was appropriate for transplant,
0:21:11 where do we manage them in the hospitals?
0:21:13 Ultimately, we went to the home.
0:21:14 How are we gonna do that?
0:21:15 What’s the actual therapy as well?
0:21:17 Supportive therapies are going around,
0:21:19 including supportive therapy for the family.
0:21:22 And we slowly but surely worked our way through
0:21:25 not only the individual treatment,
0:21:27 but also all the processes that we needed to go along.
0:21:29 And I think what you’re gonna see in, for example,
0:21:31 cell therapy in many of these other places
0:21:34 is that not only do we develop in-of-one activities
0:21:37 where we say biological plausibility and safety
0:21:41 and good manufacturing and, you know, effectiveness,
0:21:44 but we also start to refine how they perform
0:21:48 in a greater scenario or greater system of care.
0:21:51 And we’ve seen that happen over and over again across time.
0:21:53 – So, you know, getting back to this genia of the bottle,
0:21:56 the fact that you can get these CRISPR kits, you know,
0:21:59 on Amazon, and actually literally you can YouTube this,
0:22:01 there’s like some guy that has like 15 glowing dogs.
0:22:05 You know, what, how do you think about that
0:22:07 when people can do things like that, you know,
0:22:07 in their home?
0:22:10 How do you think about what should the FDA do
0:22:12 about things like that?
0:22:15 – So, you know, it goes back to this issue of authorities
0:22:18 and sort of where’s our, you know, core responsibility
0:22:20 and how we need to move things forward.
0:22:23 You know, practically speaking, we’re responsible
0:22:25 for thinking about medical products
0:22:29 that are now going into commercial, commercial use,
0:22:33 and specifically for now the treatment
0:22:37 of medical problems or sort of justifiable claims.
0:22:40 So, the individual patient who’s buying it off the internet,
0:22:42 injecting it into their side,
0:22:44 it gets into this very fuzzy area
0:22:46 of exactly what are our authorities.
0:22:49 And I think that becomes now really much more
0:22:52 of a national conversation, a route, rights, and privacy,
0:22:55 as opposed to, you know, FDA approval of the kit.
0:22:58 And practically speaking, if the kit was gonna be approved
0:23:02 for commercial purposes with claims and labeling, et cetera,
0:23:05 that’s when it starts to get into the FDA perspective.
0:23:09 It gets really murky when we live in this landscape
0:23:11 of the internet without claims.
0:23:14 You know, we see that pop up, not just in CRISPR kits,
0:23:16 but CBD and vaping.
0:23:18 Like, there’s a lot of other places.
0:23:20 And, you know, that’s why it kind of went back
0:23:21 to that point around authorities.
0:23:23 Like, there’s sort of like really clear guidelines
0:23:24 of what does the law say?
0:23:27 And then like as you move out, how do we think about that?
0:23:30 – So, and, you know, maybe the ultimate sci-fi example
0:23:34 of thinking about FDA into new areas are that, you know,
0:23:36 even algorithms themselves, you know,
0:23:38 can have therapeutic or diagnostic value.
0:23:42 And, you know, how do you think about sort of regulating
0:23:44 these algorithms themselves as they change
0:23:47 and go through revisions and have impact
0:23:49 on how we make these either clinical decisions
0:23:51 or even our therapies themselves?
0:23:53 – So this is a really important area of, again,
0:23:56 new regulatory paradigms and really trying to figure out,
0:23:59 like, how do we do this?
0:24:03 And as I think about algorithms and the
0:24:06 regulatory paradigms around them, first of all,
0:24:08 I tend to divide this into two main categories.
0:24:11 Algorithms that have a responsibility
0:24:12 of acting as a medical treatment.
0:24:14 So essentially software is a medical device.
0:24:16 And there, there’s a risk-based paradigm
0:24:19 that asks the question, does this particular
0:24:24 software product ultimately basically take the place
0:24:26 of the judgment of the physician
0:24:28 and ultimately now make a clinical decision
0:24:31 on the physician’s behalf without the physician intervening?
0:24:33 And depending on whether or not that’s gonna happen,
0:24:37 then there’s a differing set of expectations
0:24:41 in terms of the development of the regulatory paradigm.
0:24:42 So a couple of issues, though, that goes along with this.
0:24:47 One is that as software can update so quickly,
0:24:51 developing regulatory paradigms that allow also update cycles
0:24:53 that keep pace with software update cycles
0:24:56 is one of the things that, as FDA, we’re working on.
0:24:57 – Yeah, actually, is that even possible?
0:24:59 I mean, ’cause people can update software
0:25:00 obviously very quickly.
0:25:04 – So this is something that, as FDA,
0:25:07 we’ve been speaking publicly about quite a bit.
0:25:11 Can we come up with essentially preconditions
0:25:14 for software updates so that if there are
0:25:17 strong quality controls in the way software is developed,
0:25:22 well-understood product performance in terms
0:25:23 of the expectation of the updates,
0:25:28 can you now have algorithm updates that are as expected
0:25:30 and don’t require the same level of review?
0:25:32 And so that’s something that we’re certainly
0:25:34 spending a lot of time working on
0:25:36 through a series of pilot projects.
0:25:38 Also, the other part of what you’ve just mentioned
0:25:42 in terms of our sci-fi land is that, practically speaking,
0:25:46 software and algorithms are actually also innovating
0:25:51 all across the spectrum of life sciences and healthcare
0:25:53 just outside of what we regulate.
0:25:55 So it may not necessarily be a software product
0:26:00 that’s acting as a diagnostic or treatment activity,
0:26:03 but it’s a software product that’s intended
0:26:05 to support life sciences more globally,
0:26:08 whether that’s to make clinical trials more efficient,
0:26:09 to match patients to clinical trials,
0:26:13 to curate data, to help do workflow in the hospital.
0:26:15 And all of those kinds of software products,
0:26:19 we don’t directly regulate, but importantly,
0:26:21 those products also need some good signals
0:26:23 of here’s what good looks like,
0:26:25 and here’s how you should think about good software controls
0:26:26 in those settings as well.
0:26:28 – Actually, one bit of news that came up
0:26:32 was Google purchasing a huge amount of healthcare data.
0:26:36 And data and understanding how that gets regulated
0:26:40 is I would think would also be a really tough question.
0:26:44 I mean, how do you think about how these new kinds of data
0:26:45 that people are generating,
0:26:48 and then new people who want to get access to that,
0:26:50 how does they have to think about ownership
0:26:53 of the data, privacy, and what are the opportunities
0:26:54 there and the challenges?
0:26:56 – I thought you were gonna go there in this session.
0:26:57 (laughing)
0:27:01 So data ownership and privacy.
0:27:03 So there’s an easy way for me to get out of this
0:27:05 is the FDA, which is the practically speaking,
0:27:07 when the data comes to the FDA.
0:27:10 It’s the, much of the data that comes to the FDA
0:27:12 is the proprietary information and confidential information
0:27:14 that belongs to the company.
0:27:16 And so we treat it as confidential information.
0:27:18 And then there’s other information that we use,
0:27:21 for example, for drug surveillance and those kind of things
0:27:23 that are sort of more publicly available data sets.
0:27:28 So, in a lot of ways, I think that the easy FDA answer
0:27:30 is we don’t have a lot of things
0:27:32 that we specifically have to worry about.
0:27:35 But in my CIO role, I just recently started pushing
0:27:37 on the fact that I really think we need
0:27:39 a chief privacy role at FDA,
0:27:41 which we don’t currently have.
0:27:44 And when I brought this up, people said,
0:27:48 well, the data that comes to us is de-identified
0:27:51 where this is not the problem that we’re really living in.
0:27:54 But if we go back to what prompted the question,
0:27:57 which is with the Ascension data
0:28:00 and what’s going on right now in the Google story,
0:28:05 I think practically speaking, our laws of the past, HIPAA,
0:28:07 really contemplated a different world
0:28:08 than we live in right now.
0:28:12 Whereby, essentially in 2019 and going forward,
0:28:16 it’s very hard to maintain privacy of any individual.
0:28:18 – It may be even hard to denominize.
0:28:19 – It’s really very hard to denominize.
0:28:22 And it’s not just genomics data.
0:28:24 The launch to the whole story of your healthcare,
0:28:26 every single time you visited the doctor,
0:28:30 how much medicine you received, whether or not
0:28:33 you got an additional test such as an EKG,
0:28:36 that pattern is your unique footprint as well.
0:28:38 And so there’s lots of different ways
0:28:42 that data these days actually has a unique
0:28:45 and representative pattern that really is individual to you.
0:28:48 And I think the reason I brought this up at FDA
0:28:52 as a Chief Privacy Officer is that I think practically speaking,
0:28:54 even information that is de-identified
0:28:56 from a HIPAA perspective actually still
0:28:59 is probably re-identifiable even in our context.
0:29:00 And we need to be starting to think about
0:29:02 what does that mean now and in the future.
0:29:05 And also, what are some of the creative ways
0:29:06 that we can start to prepare?
0:29:08 Some of it’s just having the conversation.
0:29:13 Some of it is making sure that we are absolutely fierce
0:29:16 when it comes to security and understanding
0:29:19 who’s accessing what data for what purposes.
0:29:20 But it’s also, when do we start using,
0:29:21 for example, synthetic data?
0:29:23 How do we actually start to think about
0:29:26 what new tools and techniques and tricks we can use
0:29:29 for data in the out and in the future to preserve privacy?
0:29:32 And I think it’s all of our responsibility.
0:29:33 – Well, so I also wanted to talk to you
0:29:34 about a different way to think about data,
0:29:37 which is we could imagine in clinical trial
0:29:39 the future that’s maybe fairly different.
0:29:43 So, I think, I don’t think anybody disagrees
0:29:46 or maybe I could be wrong that it’s important
0:29:48 for the FDA to test toxicity.
0:29:50 We don’t wanna put out things that are toxic.
0:29:53 But maybe a bold thing to say is that FDA
0:29:55 will run phase one clinical trials,
0:29:57 will review phase one clinical trials,
0:29:59 but maybe we don’t need phase two or three.
0:30:03 Maybe, especially for maybe life-threatening diseases,
0:30:07 we let real world evidence and payers decide efficacy
0:30:09 as they’re gonna do anyways through reimbursement.
0:30:11 And that maybe the FDA could actually pull back
0:30:13 and data gets used differently.
0:30:15 I think, do you think that’s viable?
0:30:19 – So, three things to kind of go into this
0:30:21 that will underscore ultimately what I believe
0:30:22 is a resounding yes.
0:30:25 So, practically speaking, we are already starting
0:30:30 to see new drug development paradigms in terms of
0:30:32 starting to shake up their traditional phase one,
0:30:35 phase two, phase three happen.
0:30:39 And practically speaking, we’ve seen drugs approved
0:30:40 based on phase one data.
0:30:42 We’ve seen expansion cohorts all happen
0:30:43 within the phase one setting,
0:30:46 which is basically now end up with phase one trials
0:30:47 with a thousand patients on a phase one trial.
0:30:50 That’s not the way I was taught as a clinical trial.
0:30:51 – And they’re sort of phase one slash two
0:30:52 or something like that.
0:30:53 – Yeah, there’s some of them phase one, too.
0:30:55 For them, they’re just expansion cohorts
0:30:57 in what’s traditionally caused phase one.
0:31:00 But practically speaking, I think that what we’re seeing
0:31:01 is a blurring of the phases.
0:31:02 What we’re also seeing now is
0:31:05 contemplation of platform trials.
0:31:07 We’ve been talking about platform trials for a while.
0:31:08 They’re really hard to pull off.
0:31:11 They’re hard to pull off because of issues
0:31:13 of contracting and intellectual property.
0:31:14 They’re hard to pull off
0:31:16 because the underlying infrastructure is tough,
0:31:18 which I’ll come back to.
0:31:21 But practically speaking, we’ve talked about platform trials
0:31:24 where we can now, in one clinical trial setting,
0:31:26 start to evaluate multiple investigational products
0:31:28 simultaneously.
0:31:33 – We’ve started to say, once approving a drug,
0:31:38 start to now use information in the real world setting,
0:31:40 whether that is prospective or retrospective,
0:31:42 but classically said, real world data
0:31:44 and real world evidence to start to create
0:31:47 a total product story or a totality of the evidence
0:31:49 around this particular product.
0:31:51 So I think this is the landscape we’re going to.
0:31:54 I also think that we had essentially
0:31:59 an accelerant put into the story in December 2016,
0:32:03 which was 21st century cures.
0:32:04 If you look underneath the hood
0:32:06 of the 21st century cures legislation,
0:32:09 what you see are a number of elements
0:32:11 that push us in the direction of starting
0:32:14 to accelerate our clinical evidence development process.
0:32:15 What kind of elements?
0:32:19 So this includes starting to double down
0:32:21 on how we think about surrogate endpoints,
0:32:24 how we use patient-reported data in the process,
0:32:26 how we actually start to understand
0:32:28 and enable platform trials,
0:32:29 how we now use real world evidence,
0:32:33 and asking FDA to get smart about when
0:32:35 and we can confidently use real world evidence.
0:32:38 And so I think that all of those were enabling features
0:32:40 within 21st century cures,
0:32:41 and now we all have a responsibility
0:32:43 to start to figure it out.
0:32:45 My last point around this,
0:32:47 which is that it is really hard to do
0:32:50 because it takes putting the toe in the water,
0:32:53 and that has to happen with some company
0:32:58 or some investigators core baby.
0:32:59 And that actually is really hard
0:33:02 because it’s hard to want to subject
0:33:04 your particular product that you’re studying right now.
0:33:06 It may be your only shot on goal
0:33:08 into a clinical evidence framework
0:33:10 that we’re still trying to all figure out.
0:33:12 And so I’m not surprised it’s taking us a while
0:33:13 to figure out.
0:33:16 We have to figure out not only how to do the work
0:33:19 of new clinical evidence development paradigms,
0:33:21 but actually people have to be ready to participate
0:33:24 and it’s taking us a while.
0:33:25 – You know, and it’s interesting you mentioned
0:33:26 21st century cures.
0:33:28 You know, I’m curious about, you know,
0:33:30 to connect this to how we think about
0:33:32 how innovations like this happen
0:33:34 with all the political landscape
0:33:35 that has to make it happen.
0:33:37 And, you know, what is this interplay
0:33:40 between politics and the FDA?
0:33:41 I mean, I hear there’s an election,
0:33:42 you know, coming up sometime soon,
0:33:44 and that these are things that are,
0:33:47 you know, these things turn into realities
0:33:50 for the life that we all have to deal with here.
0:33:51 – It’s really interesting.
0:33:53 So if we look back,
0:33:56 so Medicare Modernization Act was passed in 2003.
0:33:57 So I’ll sort of use that as my starting point.
0:34:00 If I look back to MMA in 2003,
0:34:02 around that time,
0:34:07 we were contemplating sort of new payment delivery models,
0:34:09 comparative effectiveness.
0:34:12 There was a report from the Institute of Medicine in 2007
0:34:14 around building a learning healthcare system
0:34:17 where basically available interconnected data
0:34:19 could help us continuously optimize
0:34:21 both healthcare delivery,
0:34:25 but also understanding performance of drugs and devices.
0:34:30 That piece of work from the Institute of Medicine
0:34:33 in 2007 basically said in order to pull this off,
0:34:35 we need a digital infrastructure.
0:34:37 And basically, you know, it was a treatise,
0:34:39 said here’s what’s going to happen.
0:34:41 The reason that’s so important
0:34:43 is that then we had something really important
0:34:46 in 2008 or so, the global financial crisis,
0:34:48 which then led to the stimulus bill.
0:34:51 So it was because that treatise was already ready
0:34:54 and also came along with the point of view
0:34:57 of we need a digital infrastructure to pull this off
0:34:59 that embedded within the context of the stimulus bill,
0:35:00 we got the high tech act,
0:35:03 which led to the full scale distribution
0:35:05 of electronic health records.
0:35:07 And we can all talk about electronic health records
0:35:09 and good and bad points of view,
0:35:14 but what you can see is that a big international experience,
0:35:18 the GFC actually then had a direct day-to-day result
0:35:20 in terms of an enabling digital infrastructure
0:35:23 in the United States circa 2009.
0:35:27 There’s a bunch of different examples along the way,
0:35:30 but if I then think about what was happening
0:35:32 in terms of parallel legislation
0:35:34 on the House and the Senate side
0:35:36 that ultimately became 21st century cures,
0:35:39 we had this conversation going on
0:35:41 around innovative legislation
0:35:44 to try and accelerate the development of cures.
0:35:46 But I don’t know if many of you probably remember it,
0:35:48 largely got put on the shelf.
0:35:53 And then we were moving into the election in November 2016,
0:35:55 the election happens,
0:35:59 and now it’s a country sort of in a rather tumultuous state
0:36:02 trying to figure out what might be bipartisan.
0:36:04 And 21st century cures gets pulled back off of the shelf
0:36:07 and in December 2016 gets signed into law.
0:36:11 And so I think, again, this was a piece of legislation
0:36:15 that had been formed over the prior two years,
0:36:18 a lot like that IOM report from 2007,
0:36:19 it was generally ready to go.
0:36:20 – Yep, and then just go–
0:36:21 – And then there was an event
0:36:23 and then that’s what pushed it along.
0:36:24 – No, that’s fascinating.
0:36:26 – Okay, so let’s change the channel one more time.
0:36:28 So let’s go to Food Network.
0:36:31 So there is an F and FDA, right?
0:36:36 And the food part I think is often underappreciated,
0:36:38 and so I’m curious to dive in there.
0:36:41 And, or we could combine shows,
0:36:42 I’m talking about Star Trek on the Food Network.
0:36:45 So one of the fascinating areas that we see
0:36:48 is genetic engineering and synthetic biology
0:36:50 connecting to food.
0:36:52 And you’re seeing things like cultured meat,
0:36:54 like meat that’s never,
0:36:56 that maybe originally the DNA came from an animal,
0:36:58 but that what you get out of it
0:37:01 is like flaminion or something like that in principle.
0:37:06 And that when you start to see that being created,
0:37:07 how do you even think about that?
0:37:11 Like, and what do you worry about?
0:37:13 How do you balance this innovation
0:37:16 with making sure that we’re being safe?
0:37:18 – So, I think these innovations
0:37:20 have shaken things up a lot, right?
0:37:22 So if we think about meat,
0:37:24 there’s a clear interplay in the United States
0:37:27 between what’s the responsibilities of the FDA,
0:37:30 what’s the responsibility of the USDA?
0:37:32 And so, just in this particular space,
0:37:35 we had to start to figure out,
0:37:37 again, that language of the authorities,
0:37:40 where do we appropriately say
0:37:42 this is the part that the FDA is responsible for,
0:37:46 and sort of our unique set of science-based skills,
0:37:48 versus this is the part that the USDA sees
0:37:50 as their core responsibility,
0:37:55 trying to keep markets intact.
0:37:59 And last year, we ultimately developed agreements
0:38:04 with USDA so that the parts of the cell culture food activity
0:38:07 they’ve got to do with cell culture, for example,
0:38:10 and that part of the equation
0:38:13 ultimately became the FDA’s responsibility.
0:38:16 And then as we moved now to marketing, et cetera,
0:38:19 it became USDA, and I’m actually sure
0:38:21 where the line got drawn,
0:38:23 but it sort of reminds me of a couple of things.
0:38:25 So first of all, I go back to this point
0:38:27 of authorities and jurisdiction.
0:38:29 So as new innovations come about,
0:38:31 we have to start to figure out,
0:38:34 do we need to change the regulatory paradigm
0:38:35 to make sure it works?
0:38:38 The second thing is that we also need to think about,
0:38:41 how do we make sure consumers understand what’s going on?
0:38:43 So what does labeling look like?
0:38:44 How do we talk about this?
0:38:46 How do we have consistent language?
0:38:50 Some of you may have heard the story last year
0:38:52 around almond milk and that almonds don’t lactate.
0:38:55 Well, you know, it’s ’cause like, practically speaking,
0:38:57 what’s rice milk and almond milk and dairy milk?
0:38:59 Like, you know, how do we make sure
0:39:02 that consumers understand what this is all about?
0:39:07 And so as I think about the innovations in food,
0:39:09 I also think about what does that mean
0:39:12 in terms of the innovations in the regulatory landscape?
0:39:15 And if we don’t try and keep those two things in lockstep,
0:39:16 we gum everything up.
0:39:18 – Well, and it’s interesting because like,
0:39:21 I’m not sure there’s long lines of people protesting
0:39:22 the fact that almonds don’t lactate, right?
0:39:23 – I think there are.
0:39:25 (laughing)
0:39:27 – But yeah, it comes from an interesting
0:39:30 different set of incentives there, right?
0:39:32 And so it’s just interesting, what do you call meat?
0:39:33 What do you call milk?
0:39:35 What do you call cheese? – Right, exactly.
0:39:36 – You know, another aspect of this
0:39:38 that I think is really fascinating is just also
0:39:40 all the things you have to do with foodborne illnesses
0:39:42 and just thinking about like,
0:39:45 how does the FDA sort of wrap their heads around that
0:39:47 considering that food could be coming from anywhere?
0:39:50 And these threats are coming from anywhere.
0:39:52 – Since we’re gonna go back to CSI for a second.
0:39:55 So these days, if there’s a foodborne illness,
0:39:58 we will take the bacteria and essentially
0:40:01 do whole genome sequencing to really understand
0:40:05 the outbreak as well as which individuals who are ill,
0:40:07 are they all related to the same outbreak?
0:40:08 And for example, with listeria,
0:40:10 which particularly likes to be cold.
0:40:15 So it tends to get stuck on the nozzle in the plant.
0:40:17 And it ends up in, for example,
0:40:20 your frozen peas or other places that, you know,
0:40:24 ultimately you can trace now through whole genome sequencing
0:40:27 the fingerprint of the listeria then all the way back
0:40:32 to the individual who had the bad product
0:40:34 at their local Whole Foods, for example.
0:40:37 And so the ability to now trace that all the way through
0:40:40 is doable through modern technology.
0:40:42 And one of the things the FDA does in concert with CDC
0:40:45 and now really through international database
0:40:49 is maintain a database of all the different genomes
0:40:51 so that we can also track back and do this more quickly.
0:40:54 And so that’s kind of where things have been going
0:40:56 in the food outbreak space.
0:41:01 The other side of it is the application of technology
0:41:03 to trying to improve our ability
0:41:05 to go essentially farm to table.
0:41:08 So for example, blockchain and distributed ledger technology
0:41:12 to make sure that we can trace all the way from the farm
0:41:14 to the grocery store.
0:41:16 And one of the things we’ve been contemplating at FDA
0:41:19 is like ultimately, could you imagine the application
0:41:22 on your phone that allows you to scan peaches
0:41:26 and understand did the peaches have a full supply chain
0:41:28 that we could monitor.
0:41:30 And so these are the kinds of things
0:41:33 that are now a part of the lexicon at FDA.
0:41:36 We have a program called Smarter Food Safety
0:41:39 and that whole book of work is around thinking
0:41:42 about how do we move this field forward.
0:41:43 – That’s fascinating.
0:41:45 Okay, so we just have a few minutes left
0:41:46 and I want to take us now to the future.
0:41:49 So we started the discussion by talking about
0:41:52 the FDA being over 100 years old.
0:41:55 And I know we’ve sort of talked about the sort of challenges
0:41:57 and the work that’s been done.
0:42:00 Thinking about, let’s think about the next 100 years
0:42:01 and where does that go?
0:42:03 We’re gonna have new types of challenges.
0:42:06 One challenge maybe just to throw at you to start
0:42:09 is that we’re gonna have even just a different way
0:42:10 of thinking about disease.
0:42:14 That there’s all this science in the science of longevity
0:42:17 of just what can keep us healthier, longer.
0:42:18 Where it’s not about treating cancer,
0:42:20 it’s not about treating Alzheimer’s.
0:42:22 It’s about making sure you never get cancer
0:42:24 or you never get Alzheimer’s.
0:42:25 And all the therapeutics that would be done
0:42:29 to expand lifespan and expand healthy span.
0:42:32 That seems like just completely paradigm breaking.
0:42:34 How do you think about that?
0:42:38 – Well, so to begin with I think that it helps us
0:42:43 to distinguish between aspects of biological aberrance.
0:42:45 Where essentially biology’s gone bad
0:42:48 and we’re trying to think about treatments to fix it.
0:42:51 Versus the really difficult construct
0:42:52 that you’re talking about,
0:42:56 which is how do we essentially apply preventative approaches
0:42:59 and have the confidence that these approaches
0:43:02 are both safe and effective in a longitudinal frame
0:43:04 that really is hard to contemplate
0:43:07 ’cause we don’t know if we’ve ever gotten there.
0:43:09 It’s really hard to know that this particular treatment
0:43:12 was indeed successful for this individual.
0:43:15 So I’m curious, as you do this here,
0:43:17 this has certainly been an area of focus for you.
0:43:18 What’s your thoughts?
0:43:21 – Yeah, I think a lot of it is also having the biomarkers
0:43:24 that you can to know that to your point,
0:43:26 I think a lot of what you’ve been talking about
0:43:28 is just it’s about measurement.
0:43:29 And it’s about understanding how those measurements
0:43:32 correlate with a harm.
0:43:34 And that I think we just need to know what to measure
0:43:37 and that’s work to be done.
0:43:39 But I think that that’s something
0:43:41 that I think is a part of the science already.
0:43:46 – And I think that if I follow on that line of thinking,
0:43:48 it’s also about longitudinality.
0:43:50 It’s about saying here’s a surrogate
0:43:52 or an intermediate set of endpoints
0:43:54 that we’re going to monitor,
0:43:56 but we’re actually gonna understand longitudinally.
0:43:59 How does that then translate to what we understand
0:44:01 is happening across time?
0:44:03 Historically, the way we’ve often thought about effectiveness
0:44:05 is sort of as a fixed book of work.
0:44:07 And I think that what you’re gonna see over time
0:44:08 is we’re gonna talk more and more
0:44:10 about longitudinal performance.
0:44:12 And this is a perfect example of that.
0:44:13 – And so one last really quick question.
0:44:16 So what does this mean for the next 100 years of the FDA?
0:44:20 What do you see, what’s your vision for it?
0:44:22 What are we gonna be talking about 100 years from now?
0:44:24 – So I think the FDA of the future
0:44:27 is gonna be far more digital and informed by data
0:44:28 at all times.
0:44:31 A lot of the activities are gonna be automated
0:44:32 so that we can focus our time and attention
0:44:34 on the things that need to happen first.
0:44:38 That we’re going to be able to ultimately understand
0:44:41 how products are performing across time
0:44:44 and actually use that information from across time
0:44:46 to right size indications in a smart way.
0:44:47 – Okay, well thank you so much.
0:44:48 – Thanks.
0:44:51 (audience applauding)
0:44:54 (audience applauding)
0:00:04 I’m Hannah.
0:00:06 The federal agency known as the FDA,
0:00:07 or the Food and Drug Administration,
0:00:09 was born over 100 years ago
0:00:12 at the turn of the Industrial Revolution
0:00:14 in a time of enormous upheaval and change
0:00:16 and rapidly emerging technology.
0:00:19 All of those things could be said to be just as true today.
0:00:23 From CRISPR to synthetic biology to using AI and medicine,
0:00:24 our healthcare system is undergoing
0:00:27 massive amounts of innovation and change.
0:00:28 This wide-ranging conversation
0:00:31 between Principal Commissioner of the FDA,
0:00:32 Amy Abernathy, and Vijay Pandey,
0:00:34 general partner at A16Z,
0:00:38 took place at A16Z’s annual summit in 2019,
0:00:41 and covers everything from gene editing your dog
0:00:43 to tracking the next foodborne outbreak,
0:00:45 how advances in bioengineering
0:00:47 are transforming healthcare, clinical trials,
0:00:48 and drug development,
0:00:50 and how the federal agency is evolving
0:00:53 to keep pace with the scientific breakthroughs coming
0:00:55 while staying true to its core mission
0:00:58 of assessing safety and effectiveness for consumers
0:01:00 in the world of food and medicine.
0:01:02 – So thank you so much for joining us.
0:01:04 – Terrific to be here, hello.
0:01:06 – So, you know, in thinking about how I start this,
0:01:08 I was thinking about the origins of the FDA.
0:01:13 So the FDA started in 1906, 113 years ago.
0:01:17 And it’s interesting to think about that time
0:01:20 because, you know, it’s turn of the previous century,
0:01:22 a time of a lot of tumult, innovation,
0:01:26 technical change, an Industrial Revolution,
0:01:28 you know, things that actually really were
0:01:31 the driving forces to create the FDA.
0:01:33 And like, here we are, another turn of the century,
0:01:34 another Industrial Revolution,
0:01:36 another amount of tumultuous change.
0:01:39 You know, what are the needs of the FDA right now?
0:01:42 And, you know, is its core mission really still relevant?
0:01:44 – Is the FDA still relevant?
0:01:45 – Yeah.
0:01:48 – So, I’m gonna go back that 113 years
0:01:50 and the time the FDA was formed
0:01:53 and continues to be the largest consumer protection agency,
0:01:55 was formed out of a hundred laws.
0:01:57 I think that the issue that was going on at the time
0:02:00 was unhygienic conditions in the Chicago stock yard.
0:02:03 And you can imagine, there’s been a lot of responsibilities
0:02:06 of the FDA over time, phyllidimide, et cetera.
0:02:09 But practically speaking, the FDA is responsible
0:02:12 as a science-based agency to protect
0:02:15 and promote public health, including through making sure
0:02:18 that we have safe and effective medical products
0:02:20 to use every day with our patients,
0:02:22 as well as through promoting innovation.
0:02:24 – You know, your question was
0:02:26 whether or not the FDA is still relevant.
0:02:27 – Yes.
0:02:28 – And I would argue that in a time
0:02:31 of rapidly emerging biology,
0:02:34 when we’ve got more and more scientific innovations
0:02:37 and potential products coming to bear,
0:02:40 the need to make sure that we have an objective way
0:02:42 of assessing safety and effectiveness
0:02:45 and providing consumer confidence
0:02:48 that this treatment is appropriate for me,
0:02:50 it is a responsibility of the FDA
0:02:52 that actually has more responsibility, not less.
0:02:53 – Well, so, you know, in that context,
0:02:55 let’s talk about what the FDA looks like today.
0:02:57 In those older days, you know,
0:03:00 the data came to the FDA by the truckload.
0:03:02 You know, I’m just imagining like,
0:03:03 reams and reams and paper and so on.
0:03:05 And you know, and I’m just curious to get your take
0:03:08 from, you know, what does the current system look like?
0:03:10 And you know, could you take us through, you know,
0:03:11 how this works?
0:03:12 – So, you know, a couple of things.
0:03:13 I actually think that there was a time
0:03:16 when it probably came on horse and buggy, not just trucks.
0:03:17 – Yes, yes, yes.
0:03:19 – So, practically speaking,
0:03:21 I usually think about five key elements
0:03:24 in drug and biologic product development.
0:03:26 So, there’s the discovery phase,
0:03:28 then there’s time of preclinical development.
0:03:31 After you’ve done adequate preclinical development
0:03:33 in line with good laboratory practice,
0:03:36 then you’d submit an investigational new drug application
0:03:39 for a drug or a biologic to the FDA,
0:03:41 which gives permission then to start
0:03:44 clinical studies with people.
0:03:46 And a drug or biologic will go through
0:03:47 a series of clinical studies,
0:03:49 typically phase one through three,
0:03:51 although these days those lines are blurring
0:03:56 for exactly what drugs follow then by a new drug application
0:03:59 or biologics application of BLA to the FDA.
0:04:02 And then the final stage after FDA marketing approval
0:04:03 would be post-marketing assessment
0:04:05 and continuous surveillance
0:04:07 about this particular medical product.
0:04:09 So, that’s kind of the usual steps.
0:04:10 – Given that, you know,
0:04:13 what would you want to modernize about it?
0:04:16 You know, how do you take us into the next 100 years?
0:04:18 – Yeah, so, you know, I think that goes back
0:04:19 to the horse and buggy and the trucks
0:04:22 and how information has historically gotten to the FDA.
0:04:26 So, I came from the landscape
0:04:29 of a health tech startup company focused on data.
0:04:32 I was recruited to FDA with the expectation
0:04:35 that I was gonna show up and focus specifically
0:04:38 on digital innovation for the FDA
0:04:43 and sort of went smashing into a realization
0:04:45 that in fact, the way the majority of our applications
0:04:49 still come in is through PDFs
0:04:53 or sort of essentially large digital representations
0:04:56 of what used to come in on trucks.
0:04:59 We now receive most of our drug applications,
0:05:03 but not all in some kind of electronic format.
0:05:05 As a matter of fact, just two weeks ago,
0:05:09 I had to approve a book of work in orphan diseases
0:05:11 where we’re still getting things on paper as one example.
0:05:15 But most things come in as a digital application
0:05:18 also with some digital data that gets stored at FDA.
0:05:22 And as I think about where we’re going in the future,
0:05:26 practically speaking, in order to become more efficient FDA,
0:05:28 we’re gonna need to receive more and more
0:05:31 of those applications in digital formats
0:05:33 that start to represent structured data,
0:05:36 structured data that we can review at scale
0:05:38 and also structured data that allows us
0:05:41 to now continuously surveil medical products
0:05:42 in a better way in the future.
0:05:45 And so what I realized was that if I was going to be
0:05:47 a person focused on data at FDA,
0:05:50 I was gonna need to also think about how do we modernize
0:05:53 the FDA’s underlying infrastructure to take that forward.
0:05:55 And that’s why I took on the CIO job.
0:05:56 – Yeah, in addition to the infrastructure,
0:05:59 it’s interesting to think about the culture.
0:06:03 And because for example, if there’s a great drug
0:06:07 that nobody ever gets, nobody ever knows about it.
0:06:08 Let’s say there was a cure to cancer
0:06:10 but the FDA didn’t approve it.
0:06:13 There’s no outcry because no one ever knew about it.
0:06:16 But on the other hand, if the FDA lets something through
0:06:18 that actually has harmful effects,
0:06:20 thalidomide and other classic examples,
0:06:23 then there’s a huge backlash.
0:06:26 How do you build and how do you innovate in a culture
0:06:29 that has to deal with such strong asymmetries?
0:06:32 – So it’s interesting as you describe those asymmetries,
0:06:35 what you all heard describing is the practical reality
0:06:38 that that can make you very risk adverse, right?
0:06:39 – Yes, exactly.
0:06:41 I’m really worried that I’m gonna do something wrong
0:06:43 and I’m gonna flub up and that’s actually gonna have
0:06:47 very public impact and we see that all the time.
0:06:52 I think that in order to develop solutions
0:06:55 for regulatory innovation, then what you really have to do
0:06:57 is come up with flexible mechanisms
0:07:00 that also allow us to deal with the risks
0:07:03 but also take some risks when appropriate.
0:07:06 And so that means that as FDA,
0:07:08 one of the things that we focus on
0:07:11 is risk-based scientific decision-making
0:07:15 and trying to right-size the degree of review
0:07:18 and expectation with the potential risk
0:07:20 of this particular product.
0:07:22 And what do I mean by risk?
0:07:23 Sometimes there’s safety risks, right?
0:07:26 So the risk of, for example, hepatic failure
0:07:29 or the risk that the drug might take a person’s life.
0:07:32 Sometimes the risks of the size of the population impacted.
0:07:34 So you’re trying to balance the urgency
0:07:36 of this particular problem sitting in front of you
0:07:40 with the number of people where this product may impact.
0:07:45 Also, there’s risks of public perception and expectation
0:07:48 and then the last sort of set of risks is
0:07:49 where can you de-risk it?
0:07:50 – Exactly.
0:07:53 – So you can de-risk it by trying to make sure
0:07:55 that, for example, preconditions are met
0:07:57 as it relates to the manufacturing process.
0:07:59 You can de-risk it by understanding
0:08:01 in a consistent way toxicity.
0:08:04 You can de-risk it by having consistent expectations
0:08:06 around clinical effectiveness.
0:08:10 – So I think what’s interesting is to think about that
0:08:12 in even the other context of what the FDA does.
0:08:15 I think people often don’t realize that the FDA
0:08:17 isn’t just about, let’s say, approving drugs.
0:08:18 We think about clinical trials.
0:08:20 There’s a lot of things that you do
0:08:23 to protect American consumers.
0:08:25 And when we’re talking about it, it’s almost like,
0:08:28 I feel like you could have a show that’s like CSI FDA
0:08:31 or something like that where you have these investigations
0:08:32 of the crisis.
0:08:35 Like, you know, like sometimes it’s slow moving crises,
0:08:36 like the opioid epidemic.
0:08:39 How do you, like for a crisis like that
0:08:43 where it slowly sneaks up on us and then it’s too late?
0:08:45 You know, how does the FDA even think about that?
0:08:48 – Practically speaking, the FDA is responsible
0:08:49 for many types of medical products.
0:08:53 And any one of those can have a crisis, I’ve discovered.
0:08:55 So we have food and drugs.
0:08:57 We have biologics and devices.
0:08:59 We have animal food and drugs.
0:09:00 We have cosmetics.
0:09:03 We have nicotine based products, vapes.
0:09:07 And so the distribution of potential crises are real.
0:09:11 And as you think about something like the opioid crisis,
0:09:16 a problem that snuck up on us in many different ways.
0:09:21 And as I step back, and I came to the FDA in March,
0:09:23 so I’ve had sort of the opportunity to watch this
0:09:26 as an insider/outsider during this period of time.
0:09:29 You know, as information starts to accumulate,
0:09:32 that says we’ve got a really big problem here.
0:09:34 That information comes from the public.
0:09:37 It comes from across different places in government.
0:09:39 And now we need to step back as a nation,
0:09:41 but as also as an agency and say, what do we formally do?
0:09:45 And so as an agency, we put in place an action plan
0:09:48 that had several parts that focus
0:09:49 on what we’re responsible for.
0:09:52 And I’m gonna come back to that key point in just a second.
0:09:55 But then also asked, how does that interdigitate
0:09:57 with all the other plans that are going on
0:10:00 across government and also across the healthcare setting
0:10:02 to try and solve for?
0:10:05 So I’ll say the last piece that I’ve found very interesting
0:10:09 since I’ve been in government is that there are very clear
0:10:11 rules of the road of our authorities.
0:10:13 It took me a while to get used to that word,
0:10:15 but the word would be authorities.
0:10:16 – Authorities, yes, yes.
0:10:19 – Yeah, so our area is a responsibility
0:10:21 and sort of legitimate jurisdiction.
0:10:25 And practically speaking, with respect to the opioid crisis,
0:10:28 we need to go back and say, whereas FDA,
0:10:32 do we have authority to try and help resolve this problem?
0:10:36 So practically speaking, we can help reduce
0:10:39 the number of opioid tablets, for example,
0:10:42 a patient has access to after back surgery or knee surgery
0:10:45 in order to reduce the chance that this particular person
0:10:47 has access and becomes addicted in the first place.
0:10:51 As a second example, we can increase methods for access
0:10:55 for example, naloxone-based treatments in the field.
0:10:58 And we’ve done a number of projects to try and make sure
0:11:01 that there is patient-informed labeling
0:11:03 and other aspects in the field.
0:11:06 And then also, we can start to think about developing
0:11:08 and helping to develop new treatments for the treatment
0:11:10 of pain as well as new treatments for the treatment
0:11:13 of addiction and start to solve a problem on that side.
0:11:15 So as FDA, we have to stick in our swim lanes,
0:11:17 but then we have to think about how that grooves
0:11:19 with everything else across government
0:11:22 so that there’s more of a nationwide approach.
0:11:27 – Yeah, in that sense of handling the authority nature
0:11:30 of things, you’ve got to make some tough decisions,
0:11:32 like even things like contaminated food
0:11:35 going across the border, like you’ve got to inspect trucks.
0:11:37 How do you know which truck to look for?
0:11:40 – Yeah, so as I got to the agency,
0:11:42 I was surprised to find not only are responsible
0:11:47 for regulating about 20% of international GDP
0:11:51 as I mentioned, that sort of is cross a broad number
0:11:53 of products and the weed that we regulate
0:11:55 is different by product.
0:11:58 So about 15% of our food is imported.
0:12:00 We need to be able to make sure that the food
0:12:02 is appropriately safe.
0:12:05 It’s appropriately labeled that it’s legitimate
0:12:06 for sale in this country.
0:12:08 And so that means that if you’re sitting
0:12:13 at the border of Mexico, we need to basically investigate
0:12:15 trucks that are coming across the border
0:12:17 and look for violative products
0:12:19 that aren’t appropriate for sale in the United States,
0:12:22 either for safety concerns or commercial concerns.
0:12:24 How do you know which truck to look at?
0:12:26 And if we don’t get that right,
0:12:29 we can basically stop traffic for miles.
0:12:33 So we have in the case of inspecting trucks,
0:12:35 we have something called the Predict Program.
0:12:39 And the Predict Program is a 10 years old rules engine
0:12:42 that’s been written by some of the different centers
0:12:45 across FDA where the rules start to predict
0:12:50 which truck is most likely to have unsafe food.
0:12:51 Now you can imagine that if we wrote those rules
0:12:55 10 years ago, they might be old rules.
0:12:58 And it’s true that we update the rules every year,
0:13:00 but we do so by hand.
0:13:02 And as I think about trying to develop
0:13:03 a more modern agency,
0:13:06 can’t we update the way that we modernize those rules?
0:13:08 And so right now we’ve got an experiment going on
0:13:11 where we’re looking at machine learning based prediction
0:13:13 of which trucks we should inspect on the border.
0:13:18 And can we now use machine learning as a way to be smarter?
0:13:21 Importantly though, going back to your point
0:13:22 about we can’t get it wrong
0:13:25 because we lose consumer and confidence when we get it wrong.
0:13:28 We can’t just say machine learning is gonna be great,
0:13:28 let’s go.
0:13:31 We actually have to thoughtfully do the experiments
0:13:35 to say if we apply a new approach over the old rules engine,
0:13:38 are we going to now be able to improve
0:13:39 our inspection on the border?
0:13:41 – Yeah, no, I think it’s fascinating to imagine
0:13:43 that there is this world where they have to use
0:13:45 doing deep learning, machine learning to be able to do this.
0:13:48 You know, I think perhaps also we underestimate
0:13:52 maybe cases where maybe you have averted crises
0:13:53 that we never heard of.
0:13:55 You know, those might be the best TV shows.
0:13:56 I mean, are there any cases like that?
0:13:57 – As you were saying this,
0:13:59 I was trying to come up with some crises I could talk about.
0:14:01 That was actually what I was thinking.
0:14:04 So, you know, here’s one that I just recently learned about.
0:14:07 I think we’re all very worried about drug shortages.
0:14:08 I’m an oncologist by background.
0:14:10 I practiced in academia.
0:14:12 I took care of adults,
0:14:16 but certainly when we think about children with leukemia,
0:14:18 one of the drugs that’s currently in shortage
0:14:21 is a leukemia drug for pediatric leukemia.
0:14:26 And we try and think through how do we help
0:14:27 avert drug shortages?
0:14:30 And in 2018, we had something north of 50 drug shortages,
0:14:34 but the little known secret is that we helped to avert
0:14:36 over 160 drug shortages.
0:14:38 That’s because practically speaking,
0:14:40 we have developed a whole staff
0:14:42 focused on drug shortages.
0:14:45 We’ve tried to start to figure out ways to predict
0:14:47 what is going to cause a drug shortage
0:14:49 and try and intervene beforehand,
0:14:51 speaking directly to manufacturers.
0:14:54 It takes sometimes a while to build that muscle, right?
0:14:56 You can imagine, we have to first understand
0:14:57 what are the causes of drug shortages
0:14:59 and how are we gonna go after them?
0:15:02 But then practically speaking, once we do so,
0:15:04 we can help to avert that crisis.
0:15:06 I see the same kind of thing right now
0:15:07 in foodborne outbreaks,
0:15:10 where we have a call every morning at 9 a.m.
0:15:12 where we’re sort of talking about the things
0:15:13 that worry us for the day.
0:15:15 I’ve officially stopped eating.
0:15:17 – Wow, that’s not good news for any of us.
0:15:20 – ‘Cause I’m getting very concerned
0:15:21 about what foodborne outbreak there’s gonna be
0:15:22 through the day.
0:15:24 But there’s sort of like this continuous sensing
0:15:26 to try and avert problems before they come.
0:15:27 – Yeah, fantastic.
0:15:29 So let’s change the channel.
0:15:30 So we’re watching CSI.
0:15:31 Let’s switch to a different show.
0:15:33 Let’s watch, let’s get into maybe something
0:15:34 more like Star Trek.
0:15:37 So let’s talk about the future,
0:15:41 because the future that’s really becoming today,
0:15:45 like stuff that I remember like five, 10 years ago,
0:15:46 were things that I thought would be sci-fi,
0:15:51 like gene therapy, gene editing, CRISPR therapies.
0:15:55 It’s kind of crazy that I was actually talking
0:15:59 with my eldest daughter and she was finding
0:16:03 that you could get kits off of Amazon to do DIY CRISPR
0:16:06 and that she could make our dog glow in the dark.
0:16:09 She liked it not to make our dog glow in the dark,
0:16:12 but I think in the end I think it was just crazy
0:16:14 that this is the world that we’re living in.
0:16:17 And so, you know, there’s two sides of this to dive into.
0:16:20 So maybe the first side is like we’ll get to kids
0:16:21 and dogs glowing in the dark in a second,
0:16:25 but like for thinking about the clinical side of this,
0:16:28 you know, how does the FDA think about something
0:16:31 like CRISPR because it’s both gene editing,
0:16:35 it’s a therapy, there’s a delivery aspect.
0:16:38 You know, does this live with the FDA?
0:16:40 Does this live with the AMA?
0:16:42 You know, how do you even start to think about people
0:16:44 throwing these crazy new developments at you
0:16:48 that could radically transform medicine and cure disease,
0:16:50 but now has to really push the paradigm
0:16:52 for the FDA in new ways?
0:16:55 – So I think one of the things to go back to
0:16:59 is this issue of risk-based regulation.
0:17:02 You know, how are we gonna start to solve for this
0:17:05 in a way that appropriately has the right regulatory paradigm
0:17:10 for this problem at hand and aligns with the level of risk?
0:17:13 And as I think about this space, you know,
0:17:15 I think that we are living in a space
0:17:17 that gets closer and closer to customized
0:17:19 or individualized therapies,
0:17:21 the landscape of the end of one,
0:17:23 and how do we make sense of that?
0:17:26 And practically speaking, in order to get there,
0:17:29 you have to have some kind of framework
0:17:31 that you apply that says,
0:17:34 all right, before we talk about CRISPR specifically
0:17:39 or anti-sense-aligonucleotides,
0:17:41 you know, before we talked about any specific thing,
0:17:43 like what’s the framework when we’re going to apply
0:17:47 and how do we do so in a risk-based way?
0:17:50 And practically speaking, that includes,
0:17:51 what do we know about safety?
0:17:54 What do we know about safety in vivo and in vitro?
0:17:56 In animals and in the petri dish,
0:17:59 what do we know about biological plausibility?
0:18:01 You know, what’s our understanding
0:18:04 from a biology perspective that this would indeed work
0:18:06 in the way that we would expect it to work?
0:18:11 What do we have in terms of a predefined set
0:18:14 of expectations in terms of clinical outcomes
0:18:18 that we can monitor in a objective way
0:18:20 to understand whether or not this intervention
0:18:22 is making the difference that we expect it to make?
0:18:25 Also, what do we need to think about with respect
0:18:26 to whether or not this is going to apply
0:18:28 just to one individual person
0:18:32 or we might now start to apply and scale this approach
0:18:34 across multiple individuals?
0:18:36 And that’s actually going to start to balance
0:18:37 how much risk we’re going to take
0:18:38 for this particular scenario.
0:18:41 What can we think about consistency and manufacturing?
0:18:44 And manufacturing I think starts to become more and more
0:18:46 of an issue across this space.
0:18:48 And then practically speaking, what are the ethics?
0:18:50 Like what’s going to happen in the clinic?
0:18:53 We may not be responsible in our authorities for ethics,
0:18:55 but I think we’re responsible for at least consciously
0:18:57 thinking about what’s going to go on.
0:19:01 So as I think about this space of incredible new therapies
0:19:06 coming forward, we all need frameworks that we can apply
0:19:09 and where every one of us can look at through a different lens
0:19:11 and say, I can understand why that’s the order
0:19:12 and that’s how we’re going to start
0:19:13 to work our way through it.
0:19:15 – Well, so let’s drill down a little deeper
0:19:18 because the end of one framework sounds like
0:19:20 mind boggling from a therapeutic point of view,
0:19:23 but maybe there’s actually a presence.
0:19:26 So think about surgeries, like heart surgeries
0:19:28 are all kind of n equals one, people are different.
0:19:33 There’s some similarities and perhaps we’re starting to look
0:19:36 at things like CAR-T, not just as actually back up.
0:19:39 So CAR-T is in this sci-fi category.
0:19:42 You take T cells out of your body, out of your blood,
0:19:46 you re-engineer them to make them sort of supercharged
0:19:48 and you put them back into the patient.
0:19:50 And the results of that are just mind boggling
0:19:53 that tumors can melt away within days
0:19:55 and people are just literally cured of cancer.
0:19:59 And so CAR-T has sort of biopharma aspect.
0:20:02 Maybe it has like a sense of doing molecular surgery,
0:20:05 you know, like in these n equals one cases.
0:20:06 So I’m curious to get your sense,
0:20:09 like n equals one is not completely unprecedented,
0:20:12 but what are the things that we need to do
0:20:14 to bring it into the future?
0:20:17 – So, you know, not only is n one not precedent,
0:20:19 we kind of go back across medicine, across time,
0:20:21 we actually really started off as in a one medicine
0:20:24 and became more quantitative across time,
0:20:25 especially as we had interventions
0:20:27 that were applicable to populations,
0:20:28 including in the million.
0:20:30 So practically speaking,
0:20:33 we do have frameworks for in-of-one therapies.
0:20:34 You mentioned surgery.
0:20:35 Another one, you know, in the landscape
0:20:37 that I came from in cancer medicine
0:20:39 was bone marrow or sim self-transplant, right?
0:20:44 These are places where we needed to first have
0:20:46 the scientific innovation that goes along
0:20:48 with biological plausibility
0:20:51 to start to figure out how we’re going to move forward
0:20:54 with new techniques and treatments about what to do,
0:20:57 but then start to apply a systematized set of expectations
0:20:59 about how to refine this and get it right.
0:21:02 So if I go back to bone marrow transplant,
0:21:05 you know, we started to develop refined processes
0:21:09 to understand which patient was appropriate for transplant,
0:21:11 where do we manage them in the hospitals?
0:21:13 Ultimately, we went to the home.
0:21:14 How are we gonna do that?
0:21:15 What’s the actual therapy as well?
0:21:17 Supportive therapies are going around,
0:21:19 including supportive therapy for the family.
0:21:22 And we slowly but surely worked our way through
0:21:25 not only the individual treatment,
0:21:27 but also all the processes that we needed to go along.
0:21:29 And I think what you’re gonna see in, for example,
0:21:31 cell therapy in many of these other places
0:21:34 is that not only do we develop in-of-one activities
0:21:37 where we say biological plausibility and safety
0:21:41 and good manufacturing and, you know, effectiveness,
0:21:44 but we also start to refine how they perform
0:21:48 in a greater scenario or greater system of care.
0:21:51 And we’ve seen that happen over and over again across time.
0:21:53 – So, you know, getting back to this genia of the bottle,
0:21:56 the fact that you can get these CRISPR kits, you know,
0:21:59 on Amazon, and actually literally you can YouTube this,
0:22:01 there’s like some guy that has like 15 glowing dogs.
0:22:05 You know, what, how do you think about that
0:22:07 when people can do things like that, you know,
0:22:07 in their home?
0:22:10 How do you think about what should the FDA do
0:22:12 about things like that?
0:22:15 – So, you know, it goes back to this issue of authorities
0:22:18 and sort of where’s our, you know, core responsibility
0:22:20 and how we need to move things forward.
0:22:23 You know, practically speaking, we’re responsible
0:22:25 for thinking about medical products
0:22:29 that are now going into commercial, commercial use,
0:22:33 and specifically for now the treatment
0:22:37 of medical problems or sort of justifiable claims.
0:22:40 So, the individual patient who’s buying it off the internet,
0:22:42 injecting it into their side,
0:22:44 it gets into this very fuzzy area
0:22:46 of exactly what are our authorities.
0:22:49 And I think that becomes now really much more
0:22:52 of a national conversation, a route, rights, and privacy,
0:22:55 as opposed to, you know, FDA approval of the kit.
0:22:58 And practically speaking, if the kit was gonna be approved
0:23:02 for commercial purposes with claims and labeling, et cetera,
0:23:05 that’s when it starts to get into the FDA perspective.
0:23:09 It gets really murky when we live in this landscape
0:23:11 of the internet without claims.
0:23:14 You know, we see that pop up, not just in CRISPR kits,
0:23:16 but CBD and vaping.
0:23:18 Like, there’s a lot of other places.
0:23:20 And, you know, that’s why it kind of went back
0:23:21 to that point around authorities.
0:23:23 Like, there’s sort of like really clear guidelines
0:23:24 of what does the law say?
0:23:27 And then like as you move out, how do we think about that?
0:23:30 – So, and, you know, maybe the ultimate sci-fi example
0:23:34 of thinking about FDA into new areas are that, you know,
0:23:36 even algorithms themselves, you know,
0:23:38 can have therapeutic or diagnostic value.
0:23:42 And, you know, how do you think about sort of regulating
0:23:44 these algorithms themselves as they change
0:23:47 and go through revisions and have impact
0:23:49 on how we make these either clinical decisions
0:23:51 or even our therapies themselves?
0:23:53 – So this is a really important area of, again,
0:23:56 new regulatory paradigms and really trying to figure out,
0:23:59 like, how do we do this?
0:24:03 And as I think about algorithms and the
0:24:06 regulatory paradigms around them, first of all,
0:24:08 I tend to divide this into two main categories.
0:24:11 Algorithms that have a responsibility
0:24:12 of acting as a medical treatment.
0:24:14 So essentially software is a medical device.
0:24:16 And there, there’s a risk-based paradigm
0:24:19 that asks the question, does this particular
0:24:24 software product ultimately basically take the place
0:24:26 of the judgment of the physician
0:24:28 and ultimately now make a clinical decision
0:24:31 on the physician’s behalf without the physician intervening?
0:24:33 And depending on whether or not that’s gonna happen,
0:24:37 then there’s a differing set of expectations
0:24:41 in terms of the development of the regulatory paradigm.
0:24:42 So a couple of issues, though, that goes along with this.
0:24:47 One is that as software can update so quickly,
0:24:51 developing regulatory paradigms that allow also update cycles
0:24:53 that keep pace with software update cycles
0:24:56 is one of the things that, as FDA, we’re working on.
0:24:57 – Yeah, actually, is that even possible?
0:24:59 I mean, ’cause people can update software
0:25:00 obviously very quickly.
0:25:04 – So this is something that, as FDA,
0:25:07 we’ve been speaking publicly about quite a bit.
0:25:11 Can we come up with essentially preconditions
0:25:14 for software updates so that if there are
0:25:17 strong quality controls in the way software is developed,
0:25:22 well-understood product performance in terms
0:25:23 of the expectation of the updates,
0:25:28 can you now have algorithm updates that are as expected
0:25:30 and don’t require the same level of review?
0:25:32 And so that’s something that we’re certainly
0:25:34 spending a lot of time working on
0:25:36 through a series of pilot projects.
0:25:38 Also, the other part of what you’ve just mentioned
0:25:42 in terms of our sci-fi land is that, practically speaking,
0:25:46 software and algorithms are actually also innovating
0:25:51 all across the spectrum of life sciences and healthcare
0:25:53 just outside of what we regulate.
0:25:55 So it may not necessarily be a software product
0:26:00 that’s acting as a diagnostic or treatment activity,
0:26:03 but it’s a software product that’s intended
0:26:05 to support life sciences more globally,
0:26:08 whether that’s to make clinical trials more efficient,
0:26:09 to match patients to clinical trials,
0:26:13 to curate data, to help do workflow in the hospital.
0:26:15 And all of those kinds of software products,
0:26:19 we don’t directly regulate, but importantly,
0:26:21 those products also need some good signals
0:26:23 of here’s what good looks like,
0:26:25 and here’s how you should think about good software controls
0:26:26 in those settings as well.
0:26:28 – Actually, one bit of news that came up
0:26:32 was Google purchasing a huge amount of healthcare data.
0:26:36 And data and understanding how that gets regulated
0:26:40 is I would think would also be a really tough question.
0:26:44 I mean, how do you think about how these new kinds of data
0:26:45 that people are generating,
0:26:48 and then new people who want to get access to that,
0:26:50 how does they have to think about ownership
0:26:53 of the data, privacy, and what are the opportunities
0:26:54 there and the challenges?
0:26:56 – I thought you were gonna go there in this session.
0:26:57 (laughing)
0:27:01 So data ownership and privacy.
0:27:03 So there’s an easy way for me to get out of this
0:27:05 is the FDA, which is the practically speaking,
0:27:07 when the data comes to the FDA.
0:27:10 It’s the, much of the data that comes to the FDA
0:27:12 is the proprietary information and confidential information
0:27:14 that belongs to the company.
0:27:16 And so we treat it as confidential information.
0:27:18 And then there’s other information that we use,
0:27:21 for example, for drug surveillance and those kind of things
0:27:23 that are sort of more publicly available data sets.
0:27:28 So, in a lot of ways, I think that the easy FDA answer
0:27:30 is we don’t have a lot of things
0:27:32 that we specifically have to worry about.
0:27:35 But in my CIO role, I just recently started pushing
0:27:37 on the fact that I really think we need
0:27:39 a chief privacy role at FDA,
0:27:41 which we don’t currently have.
0:27:44 And when I brought this up, people said,
0:27:48 well, the data that comes to us is de-identified
0:27:51 where this is not the problem that we’re really living in.
0:27:54 But if we go back to what prompted the question,
0:27:57 which is with the Ascension data
0:28:00 and what’s going on right now in the Google story,
0:28:05 I think practically speaking, our laws of the past, HIPAA,
0:28:07 really contemplated a different world
0:28:08 than we live in right now.
0:28:12 Whereby, essentially in 2019 and going forward,
0:28:16 it’s very hard to maintain privacy of any individual.
0:28:18 – It may be even hard to denominize.
0:28:19 – It’s really very hard to denominize.
0:28:22 And it’s not just genomics data.
0:28:24 The launch to the whole story of your healthcare,
0:28:26 every single time you visited the doctor,
0:28:30 how much medicine you received, whether or not
0:28:33 you got an additional test such as an EKG,
0:28:36 that pattern is your unique footprint as well.
0:28:38 And so there’s lots of different ways
0:28:42 that data these days actually has a unique
0:28:45 and representative pattern that really is individual to you.
0:28:48 And I think the reason I brought this up at FDA
0:28:52 as a Chief Privacy Officer is that I think practically speaking,
0:28:54 even information that is de-identified
0:28:56 from a HIPAA perspective actually still
0:28:59 is probably re-identifiable even in our context.
0:29:00 And we need to be starting to think about
0:29:02 what does that mean now and in the future.
0:29:05 And also, what are some of the creative ways
0:29:06 that we can start to prepare?
0:29:08 Some of it’s just having the conversation.
0:29:13 Some of it is making sure that we are absolutely fierce
0:29:16 when it comes to security and understanding
0:29:19 who’s accessing what data for what purposes.
0:29:20 But it’s also, when do we start using,
0:29:21 for example, synthetic data?
0:29:23 How do we actually start to think about
0:29:26 what new tools and techniques and tricks we can use
0:29:29 for data in the out and in the future to preserve privacy?
0:29:32 And I think it’s all of our responsibility.
0:29:33 – Well, so I also wanted to talk to you
0:29:34 about a different way to think about data,
0:29:37 which is we could imagine in clinical trial
0:29:39 the future that’s maybe fairly different.
0:29:43 So, I think, I don’t think anybody disagrees
0:29:46 or maybe I could be wrong that it’s important
0:29:48 for the FDA to test toxicity.
0:29:50 We don’t wanna put out things that are toxic.
0:29:53 But maybe a bold thing to say is that FDA
0:29:55 will run phase one clinical trials,
0:29:57 will review phase one clinical trials,
0:29:59 but maybe we don’t need phase two or three.
0:30:03 Maybe, especially for maybe life-threatening diseases,
0:30:07 we let real world evidence and payers decide efficacy
0:30:09 as they’re gonna do anyways through reimbursement.
0:30:11 And that maybe the FDA could actually pull back
0:30:13 and data gets used differently.
0:30:15 I think, do you think that’s viable?
0:30:19 – So, three things to kind of go into this
0:30:21 that will underscore ultimately what I believe
0:30:22 is a resounding yes.
0:30:25 So, practically speaking, we are already starting
0:30:30 to see new drug development paradigms in terms of
0:30:32 starting to shake up their traditional phase one,
0:30:35 phase two, phase three happen.
0:30:39 And practically speaking, we’ve seen drugs approved
0:30:40 based on phase one data.
0:30:42 We’ve seen expansion cohorts all happen
0:30:43 within the phase one setting,
0:30:46 which is basically now end up with phase one trials
0:30:47 with a thousand patients on a phase one trial.
0:30:50 That’s not the way I was taught as a clinical trial.
0:30:51 – And they’re sort of phase one slash two
0:30:52 or something like that.
0:30:53 – Yeah, there’s some of them phase one, too.
0:30:55 For them, they’re just expansion cohorts
0:30:57 in what’s traditionally caused phase one.
0:31:00 But practically speaking, I think that what we’re seeing
0:31:01 is a blurring of the phases.
0:31:02 What we’re also seeing now is
0:31:05 contemplation of platform trials.
0:31:07 We’ve been talking about platform trials for a while.
0:31:08 They’re really hard to pull off.
0:31:11 They’re hard to pull off because of issues
0:31:13 of contracting and intellectual property.
0:31:14 They’re hard to pull off
0:31:16 because the underlying infrastructure is tough,
0:31:18 which I’ll come back to.
0:31:21 But practically speaking, we’ve talked about platform trials
0:31:24 where we can now, in one clinical trial setting,
0:31:26 start to evaluate multiple investigational products
0:31:28 simultaneously.
0:31:33 – We’ve started to say, once approving a drug,
0:31:38 start to now use information in the real world setting,
0:31:40 whether that is prospective or retrospective,
0:31:42 but classically said, real world data
0:31:44 and real world evidence to start to create
0:31:47 a total product story or a totality of the evidence
0:31:49 around this particular product.
0:31:51 So I think this is the landscape we’re going to.
0:31:54 I also think that we had essentially
0:31:59 an accelerant put into the story in December 2016,
0:32:03 which was 21st century cures.
0:32:04 If you look underneath the hood
0:32:06 of the 21st century cures legislation,
0:32:09 what you see are a number of elements
0:32:11 that push us in the direction of starting
0:32:14 to accelerate our clinical evidence development process.
0:32:15 What kind of elements?
0:32:19 So this includes starting to double down
0:32:21 on how we think about surrogate endpoints,
0:32:24 how we use patient-reported data in the process,
0:32:26 how we actually start to understand
0:32:28 and enable platform trials,
0:32:29 how we now use real world evidence,
0:32:33 and asking FDA to get smart about when
0:32:35 and we can confidently use real world evidence.
0:32:38 And so I think that all of those were enabling features
0:32:40 within 21st century cures,
0:32:41 and now we all have a responsibility
0:32:43 to start to figure it out.
0:32:45 My last point around this,
0:32:47 which is that it is really hard to do
0:32:50 because it takes putting the toe in the water,
0:32:53 and that has to happen with some company
0:32:58 or some investigators core baby.
0:32:59 And that actually is really hard
0:33:02 because it’s hard to want to subject
0:33:04 your particular product that you’re studying right now.
0:33:06 It may be your only shot on goal
0:33:08 into a clinical evidence framework
0:33:10 that we’re still trying to all figure out.
0:33:12 And so I’m not surprised it’s taking us a while
0:33:13 to figure out.
0:33:16 We have to figure out not only how to do the work
0:33:19 of new clinical evidence development paradigms,
0:33:21 but actually people have to be ready to participate
0:33:24 and it’s taking us a while.
0:33:25 – You know, and it’s interesting you mentioned
0:33:26 21st century cures.
0:33:28 You know, I’m curious about, you know,
0:33:30 to connect this to how we think about
0:33:32 how innovations like this happen
0:33:34 with all the political landscape
0:33:35 that has to make it happen.
0:33:37 And, you know, what is this interplay
0:33:40 between politics and the FDA?
0:33:41 I mean, I hear there’s an election,
0:33:42 you know, coming up sometime soon,
0:33:44 and that these are things that are,
0:33:47 you know, these things turn into realities
0:33:50 for the life that we all have to deal with here.
0:33:51 – It’s really interesting.
0:33:53 So if we look back,
0:33:56 so Medicare Modernization Act was passed in 2003.
0:33:57 So I’ll sort of use that as my starting point.
0:34:00 If I look back to MMA in 2003,
0:34:02 around that time,
0:34:07 we were contemplating sort of new payment delivery models,
0:34:09 comparative effectiveness.
0:34:12 There was a report from the Institute of Medicine in 2007
0:34:14 around building a learning healthcare system
0:34:17 where basically available interconnected data
0:34:19 could help us continuously optimize
0:34:21 both healthcare delivery,
0:34:25 but also understanding performance of drugs and devices.
0:34:30 That piece of work from the Institute of Medicine
0:34:33 in 2007 basically said in order to pull this off,
0:34:35 we need a digital infrastructure.
0:34:37 And basically, you know, it was a treatise,
0:34:39 said here’s what’s going to happen.
0:34:41 The reason that’s so important
0:34:43 is that then we had something really important
0:34:46 in 2008 or so, the global financial crisis,
0:34:48 which then led to the stimulus bill.
0:34:51 So it was because that treatise was already ready
0:34:54 and also came along with the point of view
0:34:57 of we need a digital infrastructure to pull this off
0:34:59 that embedded within the context of the stimulus bill,
0:35:00 we got the high tech act,
0:35:03 which led to the full scale distribution
0:35:05 of electronic health records.
0:35:07 And we can all talk about electronic health records
0:35:09 and good and bad points of view,
0:35:14 but what you can see is that a big international experience,
0:35:18 the GFC actually then had a direct day-to-day result
0:35:20 in terms of an enabling digital infrastructure
0:35:23 in the United States circa 2009.
0:35:27 There’s a bunch of different examples along the way,
0:35:30 but if I then think about what was happening
0:35:32 in terms of parallel legislation
0:35:34 on the House and the Senate side
0:35:36 that ultimately became 21st century cures,
0:35:39 we had this conversation going on
0:35:41 around innovative legislation
0:35:44 to try and accelerate the development of cures.
0:35:46 But I don’t know if many of you probably remember it,
0:35:48 largely got put on the shelf.
0:35:53 And then we were moving into the election in November 2016,
0:35:55 the election happens,
0:35:59 and now it’s a country sort of in a rather tumultuous state
0:36:02 trying to figure out what might be bipartisan.
0:36:04 And 21st century cures gets pulled back off of the shelf
0:36:07 and in December 2016 gets signed into law.
0:36:11 And so I think, again, this was a piece of legislation
0:36:15 that had been formed over the prior two years,
0:36:18 a lot like that IOM report from 2007,
0:36:19 it was generally ready to go.
0:36:20 – Yep, and then just go–
0:36:21 – And then there was an event
0:36:23 and then that’s what pushed it along.
0:36:24 – No, that’s fascinating.
0:36:26 – Okay, so let’s change the channel one more time.
0:36:28 So let’s go to Food Network.
0:36:31 So there is an F and FDA, right?
0:36:36 And the food part I think is often underappreciated,
0:36:38 and so I’m curious to dive in there.
0:36:41 And, or we could combine shows,
0:36:42 I’m talking about Star Trek on the Food Network.
0:36:45 So one of the fascinating areas that we see
0:36:48 is genetic engineering and synthetic biology
0:36:50 connecting to food.
0:36:52 And you’re seeing things like cultured meat,
0:36:54 like meat that’s never,
0:36:56 that maybe originally the DNA came from an animal,
0:36:58 but that what you get out of it
0:37:01 is like flaminion or something like that in principle.
0:37:06 And that when you start to see that being created,
0:37:07 how do you even think about that?
0:37:11 Like, and what do you worry about?
0:37:13 How do you balance this innovation
0:37:16 with making sure that we’re being safe?
0:37:18 – So, I think these innovations
0:37:20 have shaken things up a lot, right?
0:37:22 So if we think about meat,
0:37:24 there’s a clear interplay in the United States
0:37:27 between what’s the responsibilities of the FDA,
0:37:30 what’s the responsibility of the USDA?
0:37:32 And so, just in this particular space,
0:37:35 we had to start to figure out,
0:37:37 again, that language of the authorities,
0:37:40 where do we appropriately say
0:37:42 this is the part that the FDA is responsible for,
0:37:46 and sort of our unique set of science-based skills,
0:37:48 versus this is the part that the USDA sees
0:37:50 as their core responsibility,
0:37:55 trying to keep markets intact.
0:37:59 And last year, we ultimately developed agreements
0:38:04 with USDA so that the parts of the cell culture food activity
0:38:07 they’ve got to do with cell culture, for example,
0:38:10 and that part of the equation
0:38:13 ultimately became the FDA’s responsibility.
0:38:16 And then as we moved now to marketing, et cetera,
0:38:19 it became USDA, and I’m actually sure
0:38:21 where the line got drawn,
0:38:23 but it sort of reminds me of a couple of things.
0:38:25 So first of all, I go back to this point
0:38:27 of authorities and jurisdiction.
0:38:29 So as new innovations come about,
0:38:31 we have to start to figure out,
0:38:34 do we need to change the regulatory paradigm
0:38:35 to make sure it works?
0:38:38 The second thing is that we also need to think about,
0:38:41 how do we make sure consumers understand what’s going on?
0:38:43 So what does labeling look like?
0:38:44 How do we talk about this?
0:38:46 How do we have consistent language?
0:38:50 Some of you may have heard the story last year
0:38:52 around almond milk and that almonds don’t lactate.
0:38:55 Well, you know, it’s ’cause like, practically speaking,
0:38:57 what’s rice milk and almond milk and dairy milk?
0:38:59 Like, you know, how do we make sure
0:39:02 that consumers understand what this is all about?
0:39:07 And so as I think about the innovations in food,
0:39:09 I also think about what does that mean
0:39:12 in terms of the innovations in the regulatory landscape?
0:39:15 And if we don’t try and keep those two things in lockstep,
0:39:16 we gum everything up.
0:39:18 – Well, and it’s interesting because like,
0:39:21 I’m not sure there’s long lines of people protesting
0:39:22 the fact that almonds don’t lactate, right?
0:39:23 – I think there are.
0:39:25 (laughing)
0:39:27 – But yeah, it comes from an interesting
0:39:30 different set of incentives there, right?
0:39:32 And so it’s just interesting, what do you call meat?
0:39:33 What do you call milk?
0:39:35 What do you call cheese? – Right, exactly.
0:39:36 – You know, another aspect of this
0:39:38 that I think is really fascinating is just also
0:39:40 all the things you have to do with foodborne illnesses
0:39:42 and just thinking about like,
0:39:45 how does the FDA sort of wrap their heads around that
0:39:47 considering that food could be coming from anywhere?
0:39:50 And these threats are coming from anywhere.
0:39:52 – Since we’re gonna go back to CSI for a second.
0:39:55 So these days, if there’s a foodborne illness,
0:39:58 we will take the bacteria and essentially
0:40:01 do whole genome sequencing to really understand
0:40:05 the outbreak as well as which individuals who are ill,
0:40:07 are they all related to the same outbreak?
0:40:08 And for example, with listeria,
0:40:10 which particularly likes to be cold.
0:40:15 So it tends to get stuck on the nozzle in the plant.
0:40:17 And it ends up in, for example,
0:40:20 your frozen peas or other places that, you know,
0:40:24 ultimately you can trace now through whole genome sequencing
0:40:27 the fingerprint of the listeria then all the way back
0:40:32 to the individual who had the bad product
0:40:34 at their local Whole Foods, for example.
0:40:37 And so the ability to now trace that all the way through
0:40:40 is doable through modern technology.
0:40:42 And one of the things the FDA does in concert with CDC
0:40:45 and now really through international database
0:40:49 is maintain a database of all the different genomes
0:40:51 so that we can also track back and do this more quickly.
0:40:54 And so that’s kind of where things have been going
0:40:56 in the food outbreak space.
0:41:01 The other side of it is the application of technology
0:41:03 to trying to improve our ability
0:41:05 to go essentially farm to table.
0:41:08 So for example, blockchain and distributed ledger technology
0:41:12 to make sure that we can trace all the way from the farm
0:41:14 to the grocery store.
0:41:16 And one of the things we’ve been contemplating at FDA
0:41:19 is like ultimately, could you imagine the application
0:41:22 on your phone that allows you to scan peaches
0:41:26 and understand did the peaches have a full supply chain
0:41:28 that we could monitor.
0:41:30 And so these are the kinds of things
0:41:33 that are now a part of the lexicon at FDA.
0:41:36 We have a program called Smarter Food Safety
0:41:39 and that whole book of work is around thinking
0:41:42 about how do we move this field forward.
0:41:43 – That’s fascinating.
0:41:45 Okay, so we just have a few minutes left
0:41:46 and I want to take us now to the future.
0:41:49 So we started the discussion by talking about
0:41:52 the FDA being over 100 years old.
0:41:55 And I know we’ve sort of talked about the sort of challenges
0:41:57 and the work that’s been done.
0:42:00 Thinking about, let’s think about the next 100 years
0:42:01 and where does that go?
0:42:03 We’re gonna have new types of challenges.
0:42:06 One challenge maybe just to throw at you to start
0:42:09 is that we’re gonna have even just a different way
0:42:10 of thinking about disease.
0:42:14 That there’s all this science in the science of longevity
0:42:17 of just what can keep us healthier, longer.
0:42:18 Where it’s not about treating cancer,
0:42:20 it’s not about treating Alzheimer’s.
0:42:22 It’s about making sure you never get cancer
0:42:24 or you never get Alzheimer’s.
0:42:25 And all the therapeutics that would be done
0:42:29 to expand lifespan and expand healthy span.
0:42:32 That seems like just completely paradigm breaking.
0:42:34 How do you think about that?
0:42:38 – Well, so to begin with I think that it helps us
0:42:43 to distinguish between aspects of biological aberrance.
0:42:45 Where essentially biology’s gone bad
0:42:48 and we’re trying to think about treatments to fix it.
0:42:51 Versus the really difficult construct
0:42:52 that you’re talking about,
0:42:56 which is how do we essentially apply preventative approaches
0:42:59 and have the confidence that these approaches
0:43:02 are both safe and effective in a longitudinal frame
0:43:04 that really is hard to contemplate
0:43:07 ’cause we don’t know if we’ve ever gotten there.
0:43:09 It’s really hard to know that this particular treatment
0:43:12 was indeed successful for this individual.
0:43:15 So I’m curious, as you do this here,
0:43:17 this has certainly been an area of focus for you.
0:43:18 What’s your thoughts?
0:43:21 – Yeah, I think a lot of it is also having the biomarkers
0:43:24 that you can to know that to your point,
0:43:26 I think a lot of what you’ve been talking about
0:43:28 is just it’s about measurement.
0:43:29 And it’s about understanding how those measurements
0:43:32 correlate with a harm.
0:43:34 And that I think we just need to know what to measure
0:43:37 and that’s work to be done.
0:43:39 But I think that that’s something
0:43:41 that I think is a part of the science already.
0:43:46 – And I think that if I follow on that line of thinking,
0:43:48 it’s also about longitudinality.
0:43:50 It’s about saying here’s a surrogate
0:43:52 or an intermediate set of endpoints
0:43:54 that we’re going to monitor,
0:43:56 but we’re actually gonna understand longitudinally.
0:43:59 How does that then translate to what we understand
0:44:01 is happening across time?
0:44:03 Historically, the way we’ve often thought about effectiveness
0:44:05 is sort of as a fixed book of work.
0:44:07 And I think that what you’re gonna see over time
0:44:08 is we’re gonna talk more and more
0:44:10 about longitudinal performance.
0:44:12 And this is a perfect example of that.
0:44:13 – And so one last really quick question.
0:44:16 So what does this mean for the next 100 years of the FDA?
0:44:20 What do you see, what’s your vision for it?
0:44:22 What are we gonna be talking about 100 years from now?
0:44:24 – So I think the FDA of the future
0:44:27 is gonna be far more digital and informed by data
0:44:28 at all times.
0:44:31 A lot of the activities are gonna be automated
0:44:32 so that we can focus our time and attention
0:44:34 on the things that need to happen first.
0:44:38 That we’re going to be able to ultimately understand
0:44:41 how products are performing across time
0:44:44 and actually use that information from across time
0:44:46 to right size indications in a smart way.
0:44:47 – Okay, well thank you so much.
0:44:48 – Thanks.
0:44:51 (audience applauding)
0:44:54 (audience applauding)
The federal agency known as the FDA, or the Food and Drug Administration, was born over 100 years ago—at the turn of the industrial revolution, in a time of enormous upheaval and change, and rapidly emerging technology. The same could be said to be just as true today. From CRISPR to synthetic biology to using artificial intelligence in medicine, our healthcare system is undergoing massive amounts of innovation and change.
Covering everything from gene-editing your dog to tracking the next foodborne outbreak, this wide-ranging conversation between Principal Commissioner of the FDA Amy Abernethy and Vijay Pande, GP on the Bio Fund at a16z, discusses how the agency is evolving to keep pace with the scientific breakthroughs coming, while staying true to its core mission of assessing safety and effectiveness for consumers in the world of food and medicine.
Highlights:
What the FDA looks like today and the key steps of the FDA process to getting a drug/product to market [2:20]
How to manage a culture when mitigating risk is a top priority while aiming to innovate for the future [5:22]
Creative problem-solving in times of crisis, such as the Opioid crisis [9:58]
Preparing for and preventing drug shortages at scale [13:30]
How advances in bioengineering are transforming healthcare [16:00]
How the FDA is thinking about n=1 therapies and its applications in the future [18:54]
The future of healthcare privacy [26:10]
The ways the clinical trial process are shifting [29:26]
Innovations in Bioengineering as they relate to regulating food in the future [36:02]
How the FDA handles foodborne illnesses and its plans to innovate food safety [39:12]
Discussion about the next 100 years of the FDA [41:25]