Episode Transcript
Transcripts are displayed as originally observed. Some content, including advertisements may have changed.
Use Ctrl + F to search
0:00
New customers, download the DraftKings sportsbook
0:02
app and use code defend to get
0:04
two hundred dollars in bonus bets instantly
0:06
when you place a five dollar bet on anything.
0:09
That's code defend. Only at
0:11
DraftKings sportsbook, gambling problem
0:13
called one-eight hundred gambler, twenty one- and over in
0:15
physically present in Ohio. Ballad one offer
0:17
per first time depositors who have not already redeemed
0:20
two hundred dollars and free bets via prelaunch
0:22
offer. Minimum five dollar deposit in wager. Two
0:24
hundred dollars issued as bonus bets eligibility
0:26
restrictions apply. See DKNG dot
0:28
co slash oh for terms.
0:31
Hey
0:31
y'all. I'm Sam Sanders, and I wanna tell y'all
0:33
a little bit about a new podcast that I
0:35
co host. Called Vivec. But first,
0:37
gotta introduce my cohost. Hey,
0:39
y'all. I'm Syed Jones, and I'm Zach
0:41
Stafford. On Vivec, the three of us
0:43
talk about everything. From Beyonce,
0:46
to political violence, to which
0:48
candy is the gayest
0:49
candy, more of an ass. Just just tune
0:51
in. Yeah. We literally talk about any and
0:53
everything on our show. Absolutely. Join
0:56
our group chat, come to life, follow
0:58
and listen to Vybeq wherever you get your
1:00
podcast.
1:06
I got an email from a parent of a
1:08
child. Who is
1:10
going progressively blind. This
1:12
parent writes to me and says, look, my child is
1:15
an amazing student and she's doing
1:17
so well, but she's going to lose vision.
1:20
It's going to happen. We don't know when. Can
1:22
you stop it?
1:24
Everyday, doctor Theodore EARNoff's inbox
1:26
is flooded with emails like that one,
1:29
and they put him in a difficult position
1:32
because he has the technology to
1:34
stop it Theodore is a professor
1:37
at UC Berkeley and a leading
1:39
researcher in the field of genomic therapies.
1:42
He develops medications that
1:44
change people's DNA
1:46
to cure genetic diseases, like
1:48
the one described in that email.
1:50
We are not just sitting here hand wringing at
1:53
the fault that's in our stars. We
1:55
can actually fly to the stars and touch them and
1:57
manipulate them. But having the ability
2:00
to do something and actually doing
2:02
it are two very different things.
2:08
Medicine has always suffered from a
2:10
problem called the no do
2:12
gap. It's the difference between what
2:14
we actually do for our patients and
2:16
what we could do. Given all that
2:18
we know. Breakthroughs in biomedicine
2:21
are allowing doctors to do things
2:23
they could never do before. But
2:26
sometimes these advances don't
2:28
fit into our financial or regulatory
2:30
systems. That means it can take a
2:33
long time for patients to
2:35
actually benefit, time that
2:37
many of them don't have to spare.
2:40
The National Institutes of Health invest
2:42
more than forty billion dollars in
2:44
biomedical research each year.
2:47
And the private sector in the US spends
2:49
more than twice that.
2:51
Clearly, we value these discoveries.
2:54
Why is it so hard to use them?
2:58
From the Freakonomics Radio Network, this
3:00
is FreakonomicsMD. I'm
3:02
BabuJena. Today on the show,
3:05
we'll talk about the promise of lifesaving
3:08
genetic treatments. But
3:09
first, how can we find the people
3:11
who might benefit from them? What
3:13
if artificial intelligence could examine
3:16
the fingerprints, the breadcrumbs that patients
3:18
leave throughout the healthcare system.
3:21
And feed our earn off will tell us how
3:23
editing the human genome can
3:25
cure disease and why his
3:27
answers to those desperate emails aren't
3:29
so straightforward. Our ability
3:32
to engineer these CRISPR medicines has
3:34
far outpaced
3:36
how these medicines are actually built,
3:38
tested, and put into human beings.
3:54
I'm Garv Single. I'm a physician and computer
3:56
scientist, studied artificial intelligence
3:58
and robotics, and ultimately
4:01
became a doctor to see patients. When you
4:03
say you have a background in robotics, does
4:05
that just mean you used to play with Legos Yeah. No.
4:07
Actually, I helped build a Lego
4:09
based team of autonomous soccer playing
4:11
robots as an undergraduate. And
4:13
they were in the World Cup recently? They were in
4:15
the Robo Cup. Actually, we played
4:17
at Carnegie Mellon and got
4:19
destroyed by robots that had seven
4:22
wheels. We had two wheels and, you know,
4:24
lost Are you joking or is this I'm a hundred
4:26
percent serious. When
4:28
he's not playing with Legos, my friend
4:30
Gaurav, spends his time using computer
4:33
science to solve health care problems.
4:36
Most recently, he was the chief data officer
4:38
of a company called Foundation Medicine,
4:41
which develops tests that diagnose cancer
4:43
patients with specific genetic mutations.
4:47
Now he sees patients at Brigham and Women's
4:49
Hospital in Boston and
4:51
advises other companies that are using
4:53
big data in artificial intelligence
4:56
to solve problems in medicine. Artificial
4:59
intelligence in the doctor's office may
5:02
sound as science fiction as say
5:04
soccer playing robots
5:05
but the fact is that artificial
5:08
intelligence already permeates
5:10
our lives. Credit
5:15
card companies have been using artificial intelligence
5:17
to help map risk scores. As
5:19
part of your credit evaluation, Spotify
5:22
uses art official intelligence to make personalized
5:24
recommendations, things like Google Photos
5:27
have the ability to match photos of
5:29
my children all the way from when they were born
5:31
till now when they're nine and seven, that's
5:34
incredible. The metric for a long
5:36
time has been, can computers do
5:38
things as well as humans? But
5:40
you see places like this task of
5:42
matching infant pictures to childhood pictures
5:45
where computers outperform humans. And
5:48
once you cross that threshold, you get
5:50
to real opportunity where computers could complement
5:52
humans. So when we get to medicine, I
5:54
think this becomes particularly relevant. What
5:56
are computers and artificial intelligence good at?
5:58
Two things at least. Number one,
6:01
pattern matching. Number two, doing
6:03
things very quickly. So where those two things
6:05
are important, there may be a real role for
6:07
computers and artificial intelligence. One example
6:09
where that's the case is diagnosing strokes.
6:12
When a patient has a stroke, part of
6:15
the blood flow to their brain has been blocked.
6:17
And every minute that goes by,
6:19
more and more neurons die. If you
6:21
wait too long, That brain tissue has already
6:24
died, and in fact opening up the blood vessel
6:26
no longer has any
6:27
benefit. When a patient shows up at
6:29
the emergency room with a suspected stroke,
6:31
they need to get treatment fast.
6:34
But first, to confirm what
6:36
kind of stroke they had, those
6:38
patients usually get a CT
6:40
scan. Which has to be read by
6:42
a radiologist. That's a very busy
6:44
environment.
6:45
And maybe the case that that CT scan of
6:47
the head very important, very time
6:49
sensitive, is in a queue of equally
6:51
important and equally urgent scans that
6:53
that radiologist has to read. So
6:56
it may take five minutes, ten minutes, twenty
6:58
minutes for that radiologist to review it to see if it
7:00
has a stroke, only after which can that
7:02
patient be evaluated and hopefully
7:04
treated if it's in the time window for treatment.
7:07
One place that artificial intelligence has already had an
7:09
impact is analyzing scans
7:11
of these head CTs in the emergency room
7:13
faster than radiologists ever could. One
7:16
example is a company called VIS AI that
7:18
has an FDA approved algorithm
7:20
for detecting stroke in the emergency room. It
7:22
lives on the scanner of hospitals all over the
7:24
country. I should note that this
7:27
AI is one of the companies that Gaurav
7:29
has consulted with. Every time he
7:31
scan is taken of someone's head, That
7:33
algorithm runs on that scan and
7:35
determines if it believes there's a stroke
7:38
there. If stroke is detected by
7:40
the algorithm, a radiologist immediately
7:42
reviews it determines if that patient
7:44
has indeed had a stroke and rushes
7:46
the next steps for
7:47
intervention. The result of this has been
7:49
faster detection of strokes, often by
7:51
a dozen minutes or more, stroke is
7:53
really, really common. Are there examples where
7:56
this technology is being deployed
7:58
in areas where the diseases are
8:01
much less common, what we might call rare
8:03
diseases? I
8:04
think rare diseases are the next frontier
8:06
for computational diagnostics. They're
8:09
not diseases that most providers see
8:11
every day by deaf mission and
8:13
as a result believed to be highly
8:15
underdiagnosed. Meaning, for
8:17
some of these conditions, the subset
8:19
of people who know that they have the condition
8:22
is a small minority of the people who actually
8:24
have the condition. A common expression that
8:26
we probably both heard in training is when
8:28
you hear hoof beats, think horses,
8:31
not zebras that the rare diseases one
8:33
thinks about later. On the flip
8:35
side of that, If you're a patient with a
8:37
rare disease, that can be a very frustrating
8:39
experience. It can mean going from
8:41
doctor to doctor from proposed diagnosis
8:44
to proposed diagnosis alitany
8:46
of tests and evaluations and treatments,
8:49
all without any benefit while you're
8:51
on what is often termed a diagnostic odyssey
8:54
That can take months, it can take years, it can
8:56
take lots of expense and heartache
8:58
and frustration. If there were
9:00
a way to make that diagnosis earlier,
9:03
That could be tremendously beneficial to the
9:05
patient, to the health system, avoiding
9:07
all this unnecessary work and getting on the right treatment
9:10
sooner. The
9:13
US government defines rare diseases
9:15
as those that affect fewer than two hundred
9:18
thousand people in the country. Some
9:20
affect only a handful of people. But
9:23
the word rare can be misleading when
9:25
talking about rare diseases because
9:27
there are more than seven thousand of them.
9:30
Taken altogether, more than thirty
9:32
million people in the United States
9:34
have been diagnosed with a rare disease.
9:37
That's around ten percent of the population.
9:39
So improving how we
9:41
find and care for those patients could
9:44
have a really big
9:45
impact. What if artificial intelligence
9:47
could examine the fingerprints, the breadcrumbs
9:50
that patients leave throughout the healthcare
9:52
system as part of their routine care through
9:54
the radiology imaging they've gotten, the
9:57
EKGs that have been done, the lab
9:59
work they've had done, what if it were
10:01
possible for AI to interpret
10:03
that in the background passively without
10:05
anybody even needing to think about the rare
10:07
disease and have proactively brought it up
10:10
and alert the patient or the physician that,
10:12
hey, this is something that you might want to keep an
10:14
eye out for. I noticed this pattern.
10:16
That feels like an incredibly powerful area,
10:18
and I believe there's real examples where that's now
10:24
One example might be looking at an EKG.
10:27
There's a lot more signal in EKG than
10:29
just Have you had heart attack or not?
10:31
It's incredibly rich analysis of the
10:33
electrical conduction of the heart.
10:36
And in fact, there are now a number of different
10:38
research scientists across the country that
10:40
have demonstrated that if you give
10:42
a computer sufficient training in analyzing
10:45
an EKG, Using artificial intelligence,
10:47
a computer can predict which
10:49
subset of patients is more
10:51
likely to have one of these rare conditions
10:54
Don't you worry about false positives in this sort
10:56
of scenario? That's a great point. Artificial intelligence
10:58
isn't perfect, like any screening test.
11:01
It flags people who may have
11:03
the condition, and then they can go get the
11:05
confirmatory testing. The problem is today,
11:08
we're not flagging enough people who are at
11:10
risk and as a
11:11
result, the majority of people with some of these
11:13
conditions don't know they have it. Theodore
11:15
Urnoff, the geneticist we heard from earlier,
11:18
is all too familiar with the diagnostic
11:20
odyssey that Gorev
11:22
described. It's heartbreaking. But
11:24
the good news is a key technology advance
11:26
has happened literally in the past five years
11:29
that I think is an enormous
11:31
call to action for pretty much the entirety
11:33
of the biomedical community.
11:35
Theodore isn't talking about artificial
11:37
intelligence. He's talking about
11:39
an advance in the next step on
11:41
the path to
11:42
diagnosis. The one that would come
11:44
after a patient is flagged by AI
11:47
as being at risk for rare disease.
11:50
Genetic testing. The
11:52
first complete sequence of the human genome
11:55
was obtained two decades ago. It
11:57
took about a decade in three billion dollars.
11:59
Today, there's a room on the UC Berkeley
12:01
campus and on the UCSF campus across
12:03
the bay or at Stanford. large
12:05
number of places where you can walk into the door
12:08
and a technician will take one of those little
12:10
swabs, you know, the ones you use for COVID tests
12:12
-- Mhmm. -- and swirl them in
12:14
your mouth or your nose. And twenty
12:16
four hours later, you can get a link to
12:18
your complete genetic sequence. If
12:20
you ask me when was a graduate student,
12:23
when the mid nineties at the Brown, do
12:25
you think we'll ever get to a time where
12:27
we could do that in a day, I'd go, oh, come on.
12:29
We'd more likely move faster than the speed
12:31
of light. I mean, if here we are, This isn't
12:34
some hypothetical of something that will exist in
12:36
two thousand and thirty three. This exists in
12:38
January twenty twenty three. So
12:40
the technologies to read DNA at an incredible
12:43
rate are here and they got so
12:45
much cheaper. And so
12:47
these folks who described to me
12:49
the harrowing odyssey of, like, what's wrong with
12:51
my child? Should not
12:53
suffer. We as a society. We
12:56
as a species. Owe
12:58
it to folks in such predicament to
13:01
develop and deploy a scalable
13:03
solution of rapid genetic
13:06
diagnosis.
13:07
AI could serve a lot of other functions in
13:09
healthcare, including helping
13:11
design new treatments, helping predict which treatment
13:13
is better for which
13:14
patient. But the idea of screening
13:16
for rare diseases or diagnosing undiagnosed
13:19
conditions, these feel absolutely
13:22
here and out. And as Gaurav
13:24
said earlier, artificial intelligence
13:26
is already being used in emergency
13:28
rooms to instantaneously
13:30
screen for a more common disease,
13:32
stroke. You're improving
13:35
patient's lives by decreasing morbidity
13:37
from stroke in a way that saves
13:39
payers money. And increases
13:42
the number of procedures that are done in a fee for
13:44
service environment. And so you have
13:46
incentive alignment between payers
13:48
and healthcare systems to do what's
13:50
in the best interest of patients. Implementation
13:53
is not trivial. You have to deploy software
13:55
algorithms workflow tools behind
13:57
the firewall of hundreds and thousands
14:00
of health systems. You have to get providers
14:02
to use the tools. You have to get them to make decisions
14:04
based on them. There are a lot of hurdles that
14:06
have to be overcome. And yet, when
14:08
incentives are aligned, that can happen.
14:12
The question is what will be required for the
14:14
example of detecting rare cardiovascular
14:16
condition from EKGs to become
14:18
real? The payer is caught in a bind here
14:21
because if we screen more for rare conditions,
14:23
we identify more patients who will
14:25
need to have expensive treatments. On
14:27
the other hand, you have an entire industry
14:29
that's developing novel medicines for these patients
14:32
who can't get the medicine today because they
14:34
don't even know they have the condition that's
14:36
being treated. Theodore Urnoff
14:38
is among the scientists developing those
14:41
novel medicines.
14:43
But his work comes with its own set
14:45
of economic challenges. We
14:47
owe it to the patients and the families to
14:49
aggressively
14:50
build a new framework. That's
14:53
after the break. I'm Bob Pujena,
14:55
and this is FreakonomicsMD.
15:05
This podcast is supported by Sonder
15:07
Mind. Listen, we all have
15:09
off days. But your mental health is just
15:11
as important as your physical health. If
15:13
things just don't seem to be going right,
15:15
reach out for some help. Sondermane
15:18
therapy can connect you with a therapist who takes
15:20
your insurance and is available now.
15:23
Visit sondermane dot com to meet with
15:25
a therapist who will help you get back to feeling
15:27
like, well, you. Everyone
15:29
needs a little help once in a while. So
15:32
under
15:32
mind. Therapy works. New
15:34
customers, download the DraftKings sportsbook
15:37
app and use code defend to get
15:39
two hundred dollars in bonus bets instantly
15:41
when you place a five dollar bet on anything.
15:43
That's code defend. Only at
15:46
DraftKings sportsbook, gambling called
15:48
one eight hundred gambler, twenty one and over in physicality
15:50
present in Ohio. Valed one offer per first
15:52
time depositors who have not already redeemed two
15:54
hundred dollars and free bets be a prelaunch offer.
15:57
Minimum five dollar deposit wager. Two hundred
15:59
dollars issued as bonus bets, eligibility restrictions
16:01
apply. See DKNG dot co slash
16:03
o h per terms. Hey
16:05
y'all. I'm Sam Sanders, and I wanna tell y'all
16:07
a little bit about a new podcast that I
16:09
co host called ViveCheck. But first,
16:11
gotta introduce my cohosts. Hey,
16:13
y'all. I'm sorry, Jones. And I'm the next
16:15
effort. On Vivec, the three of us
16:18
talk about everything. From Beyonce
16:20
to political violence, to which
16:22
candy is the gayest
16:23
candy, more of an ass. Just just tune
16:26
in. Yeah. We literally talk about any and
16:28
everything on our show. Absolutely. Join
16:30
our group chat, come to life, follow
16:32
and listen to Vybeq wherever you get your
16:34
podcast.
16:39
What is DNA? Explain it in a way
16:41
that someone who doesn't have a medical background
16:43
would understand. Asking a geneticist
16:45
what's DNA is like asking an
16:47
astronomer, what's a star. You
16:49
know, it's a bowl of light. Before
16:53
the break, I mentioned that doctor Fyodor
16:55
Urnoff is developing treatments
16:57
for genetic diseases. But
16:59
first, let's take step back What
17:02
is a genetic disease? And
17:04
seriously, what's DNA? Like,
17:06
what is it? Really? It's
17:09
a molecule, so it consists of atoms
17:11
just like everything else. And
17:14
it has two remarkable properties that
17:16
pretty much no other molecule has.
17:19
It can carry in it
17:21
genetic information, just like a piece of
17:23
paper, can carry a sentence.
17:26
DNA can carry genetic
17:28
sentences. In
17:29
UNI, it carries twenty thousand
17:31
such genetic constructions, which were called genes.
17:33
But the other property of DNA, which
17:36
inspired me to devote my life to it,
17:38
is that it's conceptually and
17:40
molecularly
17:41
straightforward to make a copy of it.
17:45
But that copying process isn't
17:47
foolproof. As DNA reproduces
17:50
itself again and again, sometimes
17:53
there are little typos or mutations in
17:55
the genetic instructions it
17:57
holds. That's the basis of
17:59
a process which we call evolution that
18:02
gave us this wonderful constellation
18:04
of bacteria, animals,
18:07
plants, you and I. If
18:09
DNA never changed, you
18:11
and I would still be little microbes
18:14
floating around in the primordial soup.
18:16
So this intrinsic ability
18:18
of DNA
18:19
to change is the basis of
18:22
life paradoxically. Some
18:24
of these changes have beneficial outcomes,
18:27
like a mutation that occurred around
18:29
five thousand years ago, that allowed
18:31
humans to digest lactose
18:34
for the first time. I'm happy I
18:36
inherited that one, but sometimes
18:38
the outcomes of a typo in genetic instructions
18:41
can change our lives for the worse.
18:44
Those are what we call genetic diseases.
18:47
One of the more common ones you may have heard
18:49
of is sickle cell disease. Around
18:52
one hundred thousand Americans suffer
18:54
from it. So we classify it
18:57
as a rare disease. Even
18:59
though it's the most common inherited
19:01
blood disorder in the country and
19:03
affects millions of people worldwide.
19:07
People who inherit sickle cell disease can't
19:09
form normal red blood cells that
19:11
carry oxygen. Instead, they
19:14
produce red blood cells that are rigid
19:16
and sickle shaped like a crescent. That
19:19
deformity causes extreme pain
19:22
episodes that puts sickle cell patients
19:24
in the hospital on a regular
19:26
basis. It also delays
19:28
normal development in children, damages
19:31
joints, nerves, and organs, and
19:33
often causes strokes. All
19:36
of these bad outcomes are the result
19:38
of just one letter out
19:40
of six billion in the genome
19:43
being
19:43
flipped. One tiny typos
19:45
spreads its devastating effects
19:47
through the entire body, And even with
19:50
the best healthcare, our fellow Americans
19:52
with sickle cell disease, their lifespan is
19:54
around the mid
19:54
forties. So it takes away
19:57
decades from your life.
20:01
Until now, the only way to
20:03
cure sickle cell disease was with
20:05
a bone marrow transplant. But
20:07
the procedure is not for everyone.
20:09
It can be difficult to find a well matched
20:12
donor and bone marrow transplants
20:14
are really hard on the body, especially
20:17
as patients get older. What if
20:19
instead of replacing the patients faulty
20:22
bone marrow, doctors could actually
20:24
fix the typo in the patient's
20:27
own bone marrow. That would be
20:29
a much safer procedure and
20:31
eliminate the need for a well matched donor,
20:33
meaning anyone with the disease
20:36
could potentially be cured. Thanks
20:39
to a revolutionary gene editing technology
20:41
called CRISPR, scientists like
20:44
Theodore are now doing exactly
20:46
that.
20:52
The first thing to note about CRISPR is it's one
20:54
of those acronyms. Where
20:56
what the acronym stands for is not
20:58
useful to know because it doesn't tell you anything
21:00
about what it does. And there are great examples
21:03
to the country, let's say, school, right, or
21:05
self contained underwater breathing apparatus. If
21:07
you know what the acronym stands for, you're like aha.
21:10
But CRISPR stands for,
21:11
mhmm, clustered regularly into space
21:14
short palindromic repeats, and your audience is welcome
21:16
to forget that immediately.
21:17
Or say it a hundred times, it'll help you go to sleep,
21:19
or become the least popular person
21:21
at a social gathering. Honey, I know what
21:23
CRISPR stands for. So I'm
21:26
sitting in a recording studio at
21:28
school or journalism of the University of California,
21:30
Berkeley, where I'm a professor, and
21:32
probably the single biggest discovery
21:35
in biomedicine of the past quarter
21:37
century. Was made here on the
21:39
Berkeley campus. That
21:40
discovery was made by the biochemist, Jennifer
21:43
Downner. Who, together with the Manuel
21:45
Charpentier, won the twenty
21:47
twenty Nobel Prize in Chemistry.
21:50
To be clear, Jennifer and
21:52
Emmanuel didn't create CRISPR.
21:55
CRISPR is not a fancy new lab machine.
21:58
It's a microbial defense mechanism.
22:01
It consists of just two molecules, an
22:03
enzyme that acts as a pair of DNA
22:05
scissors and a special
22:07
piece of genetic material that
22:09
tells the enzyme where in the
22:11
DNA to cut, and it's
22:14
billions of years old. Early
22:16
in the history of life, bacteria
22:18
evolved, crisper, to fight
22:20
off parasites that could attack
22:22
and kill them. It's basically little
22:25
molecular machine that carries
22:27
in it a memory of a previous
22:29
attack by a genetic invader, a
22:31
snippet of the offender's genetic material,
22:34
like a law enforcement officer with a most
22:36
wanted poster with a picture of somebody suspected
22:39
of a crime. And it literally
22:41
matches every piece of DNA it sees.
22:43
Do you have a match to this twenty
22:46
letter word that I'm carrying inside
22:48
me? If yes, I'll cut you on the tray.
22:50
If not, have a nice day. Jennifer
22:53
and Emmanuel's big discovery was
22:55
not that CRISPR exists. What
22:57
they discovered was something that CRISPR
22:59
can
22:59
do. So it turns out that you can
23:02
put CRISPR inside human cells, which
23:04
seems insane. This thing comes from
23:06
bacteria, which are billions of years apart from
23:08
us evolutionarily. You can take CRISPR,
23:10
you can give it a twenty letter match
23:13
to a human gene that's broken, and
23:15
it'll fix it. We don't understand
23:17
why it's been so successful in this
23:20
incredible new environment, but
23:22
we're grateful to mother nature
23:24
and, of course, to Jennifer and Emmanuel for
23:26
having the insight, that you can
23:28
program. This is the keyword.
23:31
CRISPR can be programmed.
23:33
Not only have scientists wielded CRISPR's
23:36
innate destructive function to
23:38
eliminate toxic genes, but
23:40
they've also come up with ways to make
23:42
CRISPR serve a constructive function.
23:45
That is to precisely alter
23:47
just one letter in the DNA to
23:50
repair a gene rather than getting rid
23:52
of it altogether. Genesis
23:54
made use of that function to
23:57
develop a cure for sickle cell
23:59
disease. Which is currently the first
24:01
CRISPR based therapeutic up for
24:03
approval by the FDA.
24:05
This is a great example of the
24:07
ways in which we humans have
24:09
borrowed from mother nature and then elaborated
24:11
on her
24:12
inventions. And we wouldn't
24:14
be talking about this if this hasn't be used on
24:16
people. CRISPR isn't the first approach
24:18
to gene therapy. There are several approved
24:21
medications that use modified viruses
24:24
to deliver disease treating genetic
24:26
material into a patient's cells.
24:28
But CRISPR cures are
24:30
the first to edit the genome
24:33
itself. So far, CRISPR
24:35
has been used to treat genetic diseases of
24:37
the bloodstream, liver, eye,
24:39
and immune system. For others,
24:42
like those affecting the lungs, brain,
24:44
and kidneys, scientists haven't
24:46
yet figured out how to get CRISPR
24:48
into enough of the organ to
24:50
actually heal it. So to be
24:53
clear, CRISPR is still far
24:55
from a cure all. But as
24:57
new techniques and technologies to deliver
24:59
CRISPR are developed, more
25:02
and more organs will come online.
25:04
Theodor expects the lungs to be next.
25:09
Having the power to cure genetic diseases
25:12
by editing the human genome is
25:14
a dream come true for geneticist. But
25:17
when it comes to using that power
25:19
to help people, the story gets
25:21
more
25:22
complicated. Many a time
25:24
when parents of children with severe genetic
25:26
disease send me an out
25:28
saying doctor Urnoff, is there anything you
25:30
can do? If they are willing to share
25:32
what the mutation is. I can load
25:35
it into some software in my computer, which is available
25:37
to all I wanna be clear on some proprietary UC
25:39
Berkeley software. And you
25:41
can basically engineer if you know what
25:43
you're doing at CRISPR to fix that mutation.
25:46
For many diseases, that Engineered
25:48
CRISPR on my computer screen can become
25:51
a vial with that CRISPR that we
25:53
can pretty quickly test for whether
25:55
it can repair the defect safely
25:57
and effectively. Start to finish.
25:59
If you know what you're doing, it'll take well under
26:01
a year. So do I write back to the
26:03
parents and say, hi, and guess what? No.
26:07
And here's why. Engineering the
26:09
medicine is the first step
26:11
of probably a four
26:13
year process to protect patients
26:15
from faulty medicines. I wanna
26:17
really emphasize, I'm not sitting here and saying,
26:20
get rid of the loss to protect patients
26:22
from faulty medicines. But going
26:24
through that four year process just to get to the clinic
26:26
takes anywhere between eight million dollars to ten million dollars
26:29
for one disease. If
26:31
the disease is relatively prevalent like sickle
26:34
cell disease and then if they charge
26:36
what is currently being charged for these types of medicines,
26:38
which is one to two to three million
26:40
dollars a patient. I can see
26:42
why a company would invest years and millions
26:44
to build a medicine. Okay. Now I get this
26:46
email from somebody and they have to children
26:48
and both children have this change and
26:50
it's unclear that anybody else on planet
26:52
Earth has that genetic
26:53
change. So who exactly is going to spend four
26:56
years and ten million dollars building a medicine
26:58
that's gonna be used to treat two kids.
27:03
Many of these diseases are individually so
27:05
rare that they do not form a
27:07
viable commercial proposition under the
27:09
current system. We need
27:12
to face the remarkable reality
27:14
that our ability to engineer these
27:16
CRISPR medicines has far
27:19
outpaced how these medicines
27:21
are actually built, tested, and put into
27:23
human beings. We have never had
27:25
a technology like CRISPR. We
27:28
owe it to the patients and the families to
27:31
aggressively build a new
27:33
framework to provide these
27:35
medicines to these
27:36
individuals. It scares me
27:38
because it's one thing to say
27:41
to somebody that we don't have a treatment
27:43
because the biology doesn't exist. To
27:45
provide that treatment. It's another thing to say that
27:47
we don't have a treatment because there's
27:49
not sufficient commercial incentive to develop
27:52
that treatment, to evaluate it, to test it.
27:54
To market it or that we
27:56
have a set of regulatory policies
27:59
that aren't adept enough to
28:01
recognize that There are some patients
28:03
with some diseases who literally
28:06
months matter in terms of getting
28:08
access to care. We want to get medicines to
28:11
people faster. But we wanna make sure that
28:13
we do so in a way that's safe. And
28:15
the FDA is really tasked with managing
28:18
that speed, safety trade off, but
28:20
Of course, that trade off should change when
28:22
the parameters change. Right? So if you have
28:24
a new technology that
28:26
will allow for personalized intervention
28:29
in people with life threatening diseases
28:31
for which early treatment
28:34
really does matter, we should be able
28:36
to create a regulatory pathway that would allow
28:38
for that. And then there's the other bucket of
28:40
our, well, how do we pay for that? That commercialization issue
28:42
is equally important. I cannot
28:44
improve on what you just say. I'll just add
28:47
one point. Many genetic
28:49
diseases are diagnosed in
28:51
human beings. At a
28:53
stage where current technology. And
28:56
I emphasize current because our
28:58
field is moving very fast. Current
29:01
technology is essentially powerless.
29:04
By the time we are looking at that human being
29:06
in a clinic, it's too late.
29:09
A really profound and poignant example
29:12
is the disease that killed Woody Guthrie,
29:14
which is Huntington's disease. It's
29:17
a broken gene. It's actually a toxic
29:19
gene, which is basically killing
29:21
the brain. And by the time people develop
29:24
symptoms, parts of their brain are
29:26
just gone, and we don't have a technology
29:28
that can bring it back.
29:30
Remember the email from the very beginning
29:32
of the episode that Theodore received
29:35
from the parent of a girl going progressively
29:37
blind. We don't know if we'll be
29:39
able to turn back time
29:41
and bring back vision. To one
29:44
hundred percent But at the very least if we
29:46
can diagnose early enough. If
29:48
we can intervene at the genetic level before
29:50
it gets worse, I can tell
29:52
you that the patient
29:53
community, the vast worldwide community of folks
29:55
with genetic disease, will applaud.
29:58
This brings us back to artificial intelligence,
30:00
and its role in catching these rare
30:02
genetic diseases as early
30:04
as possible. Here's Gaurav
30:06
Single again. Now that they're effective
30:09
treatments, it feels more important than
30:11
ever that we use techniques like this
30:13
to make sure that people who have this condition know they
30:15
have it so they can get
30:16
treated. And
30:17
in Theodore's eyes, this is a two
30:19
way street. I think we as a community
30:21
owe it to the folks
30:24
out there whose genetic changes we're identifying
30:26
is potentially dangerous or disease driving.
30:29
To make sure that our ability to address
30:31
those changes actionably in the clinic
30:33
catches up to how quickly we can identify them.
30:38
Suppose that we can solve the economic puzzle,
30:42
What's your big picture ideal
30:44
vision? A world where
30:46
genetic disease is diagnosed early
30:49
in a way that's so affordable that
30:51
health insurance just covers it. And then
30:53
in cases where that's appropriate, the
30:56
CRISPR medicine is manufactured and
30:59
administered to that individual in
31:01
a way that is scalable,
31:04
affordable, and does not involve
31:07
years and millions. Really
31:09
having, I don't know what we're gonna
31:11
call them, CRISPR clinics. Today,
31:13
I'm wearing glasses. It's a way to
31:16
correct my myopia. Do I see
31:18
a future where CRISPR is deployed to repair
31:20
genetic defects in a way that's relatively
31:22
commonplace? I do. That's
31:25
world I want to live in. So
31:28
how can we solve the economic puzzle?
31:31
How can we make developing cures
31:33
for rare genetic diseases
31:34
profitable, and accessing
31:37
those cures affordable. Financing
31:40
ends up being a tremendous roadblock but
31:42
with the right kind of financing, it actually
31:44
ends up accelerating
31:46
our ability to treat these patients. And
31:49
what's it like to receive a crisper
31:51
cure. It all amounted to a
31:53
small syringe of DNA that took
31:55
about thirty seconds to infuse.
31:58
And my life changed completely.
32:02
That's coming up next week on Freakonomics.
32:05
In the meantime, let us know what you thought
32:07
about the show. I'm at boppu at
32:09
Freakonomics dot com. That's
32:11
BAPU at freakonomix dot
32:14
com. That's it for today.
32:16
I'd like to thank my guests this week, doctor
32:18
Gaurav Singhal and doctor
32:20
Fyodor
32:20
Urnoff. And thanks to you,
32:23
of course, for listening. Fri
32:25
Freakonomics MD is part of the
32:27
Freakonomics Radio Network, which
32:30
also includes Freakonomics Radio.
32:32
No stupid question. And
32:34
people I mostly admire. All
32:37
our shows are produced by Stitcher and
32:39
Renbud Radio You can find us
32:41
on Twitter at doctor BapuPod. And
32:44
now, you can find our episodes on YouTube
32:47
too. By going to youtube dot com
32:49
slash at Freakonomics. That's
32:51
the at sign followed by freakonomics.
32:54
If you know someone who doesn't listen to
32:56
podcasts, asks, but spends a lot
32:58
of time on YouTube, let them know.
33:01
This episode was produced by Julie
33:03
Canfor and Claire Boudreaux bitch. It
33:05
was mixed by Eleanor Osbourne. Our
33:07
executive team is Neil Carruth, Gabriel
33:10
Roth, and Steven Dubner, original
33:12
music composed by Luis SCARRA. As
33:15
always, thanks for listening.
33:24
For the last twenty two years,
33:26
every time I talk to my biology
33:29
colleagues, I'm always happy
33:31
that I did a PhD in Freakonomics. But
33:34
when I'm talking to
33:35
you, I'm thinking to myself, Doug, on it,
33:37
why didn't I do something different twenty
33:40
three years ago, between you are an
33:42
MD. Right?
33:42
I'm an MD and a PG and Yeah.
33:44
Both. Wow.
33:46
That's two lives in one. Yeah. Exactly.
33:49
Yeah. I don't
33:49
know if it's a good thing or bad thing. The
33:56
Freakonomics Radio Network, the
33:58
hidden side of everything. Stitcher.
34:08
This podcast is supported by Sonder Mind.
34:10
Your mental health is just as important as your physical
34:12
health. Sonder Mind therapists can help you if
34:14
you're feeling off. They're available within days
34:17
with virtual or in person options, and
34:19
insurance is accepted. Sonder mind
34:21
dot com. Therapy works.
Podchaser is the ultimate destination for podcast data, search, and discovery. Learn More