Newborn Screening and Post-Screening Analysis of Inborn Errors of Metabolism
Articles,  Blog

Newborn Screening and Post-Screening Analysis of Inborn Errors of Metabolism


[MUSIC PLAYING] [APPLAUSE] MICHAEL GELB: Let’s see here. I’m going to start
with this short movie, but the guys remotely off
site can’t hear the sound, so I’m going to add
a little narration. You’re going to hear both. But anyways, it’s about this
remarkable new gene therapy treatment for a disease called
metachromatic leukodystrophy. It just shows where we’re going
with these remarkable therapies for these inborn errors. So basically– [VIDEO PLAYBACK] – Children with
late infantile onset metachromatic leukodystrophy
lose the ability to walk, talk, swallow, eat, and see. Most kids with MLD won’t survive
beyond their fifth birthdays. In 2010, using stem cells
derived from their patients’ blood, researchers
pioneered a treatment to repair the defective
gene that causes MLD. Giovanni price was the
second child in the world to undergo this treatment. Today, Giovanni is five years
old and has no symptoms of MLD. On August 24, 2015,
Giovanni Price did something no child with this
disease has ever done before. He walked to school for his
first day of kindergarten. [END PLAYBACK] MICHAEL GELB: So he’s
broken all the milestones, but we don’t know how
long this treatment’s going to last, right? But looks good so far. [VIDEO PLAYBACK] – By the age that
he is right now should be close to
the end of his life. And so we see the
complete opposite. I mean, we see him
going to school, and making friends, and riding
his bike in the driveway. The doctors don’t tell me
he’s, you know– they’ve never used the word cure. They’ve never– [END PLAYBACK] MICHAEL GELB: So he’s obviously
not cured, right, but– [VIDEO PLAYBACK] – I mean, this– we don’t know. But I hang on to
those words that he’s going to continue
to grow up, just like any other five-year-old. My hope is that gene
therapy combined– [END PLAYBACK] MICHAEL GELB: So gene therapy
and newborn screening– catch it early. That’s the ticket. [VIDEO PLAYBACK] – Save other families from going
through what we went through and what all of our
friends, our MLD friends go through with their kids. He has no idea that
he’s a miracle. He has no idea. [END PLAYBACK] MICHAEL GELB: Yeah,
so I’ll come back and talk about this therapy. It’s not developed here. It’s from Italy,
but it’s remarkable. I thought you should
hear about it just so you know what’s coming in
advanced kind of therapies for these diseases. My disclosure
statement, here it is. I was kind of a bad kid when I
was in high school, you know? I got in trouble
and all of that. Oh, oh, sorry. You want to hear about
my relationship with– yeah, I lied to the police
when I was 10 years old. It was a big mess, yes. My sister was there. She lied too, so I’ll
never forget that. But I grew up to be a
pretty good guy after that. So anyways, that’s
my disclosures. So newborn screening for a
treatable genetic diseases– and I have an endowed
chair position. That doesn’t mean I’m chair of
the Department of Chemistry. You would have to
torture me for 10 years to get me to agree to
become a departmental chair. So I have an endowed chair. That doesn’t mean I’m
chair of chemistry, right, just to make that clear. I don’t have any skills to be
a chairman of a major chemistry department. So what is newborn screening? I think most of
you know, but I’m going to talk about
just the gist of that and then talk about
these lysosomal storage diseases, which I think
many of you know about, but just so we’re
on the same page. And I’m going to talk
about this new technology that we’ve developed here
with Ron Scott and others on screening for these diseases. I’m going to show you
some pilot studies. I’m going to talk about
this thing of diagnosis and prognosis. You’ll see what I mean
by prognosis in a minute. And I’m going to talk about a
very exciting chapter coming as a collaboration between
me and [INAUDIBLE] at Seattle Children’s Research Institute
on newborn screening versus proteomics, which is
kind of the marriage of mass spec and newborn screening. And people like
Andy Hoofnagle are trying to apply
proteomics techniques in clinical medicine in
cases where amino acids don’t work very well. I’ll talk about that in the
context of newborn screening. So just to give you a quick
story on newborn screening, I think you know
about galactosemia, but this is lactose. And it’s not clear
why we’re born, and we eat all this lactose. It’s in the mother’s milk. It’s not the
perfect food, right, because it has one
galactose and one glucose, and yet we need 100 times more
glucose than we need galactose, right? Because glucose is our energy. It’s like the
gasoline of the car. Galactose is like
the fender of a car, the structural component. We need a lot more gasoline
than we need fenders. So the liver comes
to the rescue. And I don’t know
why we’re designing to eat food that needs to
be rescued by the liver. But I think it has something
to do with the bacteria that grow in our gut, and the
good guys live on lactose, and we want the good
guys down there. And maybe that’s it,
but nobody knows. The liver comes to
the rescue and has a three-step enzymatic
pathway to convert all of this excess
galactose to glucose. And if you have mutation in that
process and it’s inefficient, you die of galactose
poisoning, whatever that means. I suppose that means you put
galactose in places that you should put glucose. And it’s a very severe disease. And it’s fairly
simple to treat just by reducing dietary intake
of galactose, right? So we’ve been screening
for this all over the world for many years. And if we get a hit,
it’s an emergency. We call the family. Get the baby off mother’s milk. It’s total irony. You’re sort of killing your
baby with mother’s milk, cow’s milk, doesn’t matter. We got to put you on
a special formula. But if we didn’t do that– and
this is a rare disease, right? There’s very little
family history. These are autosomal
recessive diseases. Your husband and wife come from
different parts of the world, typically. They have no family history. These two come together and
have a kid with two bad genes, and you have a autosomal
recessive inborn error of metabolism. So we have to just
screen everybody in the population
in a non-biased way to find the rare case of a baby. And then we can treat it early. And if we wait until the
baby gets too sick to end up at Northgate Mall, and then
Seattle Children’s, and then another children’s. And by the time they figured it
out, the baby is too far gone and you can’t treat. So if only we had
known about this, we could have fixed it early. It’s a no brainer, right? So about 1 in 40,000
babies are born with mutations in
the DNA that leads to a defective enzyme
for converting galactose to glucose. When this happens, we get
an inborn genetic disease called galactosemia. It’s rare. It can take several
months to diagnose. I think you know all this. And by the time the
diagnosis is made, it’s too late to save the baby. If only we had
known at birth, we could have fixed it
or partially fixed it. The only reasonable solution
is to screen everybody in the population. So all sophisticated countries
around the world do this– Australia, Europe, United
States, Japan, China. And this guy, Robert Guthrie– it’s an interesting story. His niece is head of Virginia
Mason Research Institute here in Seattle, [? Jane Voigt, ?]
and I work with her husband, [? Fred Buttron ?] on parasites. So anyways, this
guy Guthrie invented the concept of newborn
screening through these dried blood spots on newborn
screening cards in the 1950s. Here’s a family
picture in Seattle. I’ve met the family
over the years. He’s no longer with us,
but he started the concept of newborn screening for
PKU, which I think you all know about, phenylketonuria, and
this idea of dried blood spots on newborn screening cards. He was at Buffalo,
University of Buffalo. So typical numbers– 1 in 100 mothers might
carry a typical mutation for a genetic
disease, let’s say, five of you in the audience. Well, the last time
I gave this talk, there were 500 people
in the audience. That was the university
faculty lecture. So let’s say two of
you in this room– what am I talking about? One of you in this
room is probably a carrier for galactosemia. And then the father comes
along, the same deal. So 1 in 100 squared
is 1 in 10,000, and 25% of the offspring
will have two bad genes. So that’s how you
get 1 in 40,000. I think you all know this. So cystic fibrosis,
right, 3,000 babies– 1 in 3,000 babies born
with cystic fibrosis, the most common
genetic disease, I think, serious genetic disease. I mean, it’s like 1 8 or
10 people are carriers. It’s amazing when
you think about it. So millions of babies
are born each year. We screen for about 40 to
50 individual disorders. According to the CDC
and American College of Medical Genetics, it’s
one of the most successful public health
programs of all times. And 1 in 300 babies, this many
per year in the United States out of 4 million
babies are saved as a result of
numerous screening, so this is a good thing. I mean, a lot of the low
hanging fruit in this area has been done. Like PKU, we should
screen for that. We have a good treatment. It’s a serious disease. Galactosemia is a no brainer. But now we’re getting into areas
that are more controversial, right, as you’ll see. Diseases that the treatments
are very expensive, the treatments
don’t cure the baby. They make the baby better,
but far from perfect. There could be late onset
variations of these diseases that scare families. It’s not so simple,
as you’ll see. Every state, more or less,
has a newborn screening lab. It’s like the state
police are involved, and when the FBI show up, the
state police say, go home. It’s our stuff. So the FBI tries to
influence the state police, but the state police doesn’t
really want the FBI there. That’s exactly how to
describe newborn screening. It’s a state by state thing. The feds try to stay out of it. But as you’ll see, it’s a
little more complicated. Every state has one. Every state has a director. Here’s John Thompson
in Shoreline runs the newborn screening
lab for Washington State. So we hear a lot about
next generation sequencing. I just want to have
one slide about this. This is a no go for first
tier newborn screening. Some genomic people
might disagree. I think they’re wrong. Here’s the deal about– pretty soon, I don’t know. Pretty soon, soon
enough we’re going to be sequencing the genome of
lots of people, lots of babies. When are we ready to use that
information for high confidence newborn screening or
prediction of disease onset? $100,000 question in this
field is, nobody knows. Is it 10 years from now? Is it 50 years from now? I don’t know. It depends on the disease. It’s a very
complicated question. It’s a very big question, OK? So the DNA sequence, if you
think it tells you everything, you’re wrong. It doesn’t. Well, first of all, it’s
still too slow and expensive for first tier population-wide
newborn screening. But that’s not
the major problem. Even if we can make it $1 per
baby, which would be the target price range, and super fast,
that’s not the major problem. If you have severe mutations,
the so-called low hanging fruit, like premature
stop codons, frame shifts, insertions, deletions, you
still have the phasing problem, right? You still don’t know if
you have both mutations on the same allele or
one from each parent. Then you’ve got to go
sequence the parent. We can do that. But there is a phasing problem. Next, large deletions
might be missed. If you have one large
deletion on one allele, not on the other, you’re just
going to read a good sequence. You’re not going
to see the allele unless you do gene counting,
which you don’t really do. You’re going to miss
large deletions. We have variations of unknown
pathogenic significance, the so-called VUS. This is not so uncommon. So if I sequence
Ron Scott’s DNA, his DNA is going to differ from
mine in one gene in about one in every 50 nucleotides. But if I look it up in a
table of 60,000 humans, I’m going to see that his
variations are well known in the population, as are mine. And so I’m going to ignore his
differences between him and me, because I see them in
so many humans that don’t have a disease. But once in a while,
not so infrequently, you find these
rare mutations that are not found in
60,000 humans, and we don’t have any
biochemical information about these missence mutations. And so they’re
officially unannotated, and they’re variations
of unknown significance. And this happens a lot. We can’t ignore these, because
we don’t know if they’re bad guys or good guys. We know they’re rare guys. The biggest problem in my mind
is what I call the ABC problem. There’s probably
another name for this. Here’s an example. A patient presents
with a genotype A/B, a set of mutations, A on one
allele and B on the other. And A and B are thought
to be partially penetrant mutations– mutations that cause
less than super severe disease, maybe later onset disease. Remember, diseases
are spectrums. So are mutations. Now we find a new patient,
B/C, also late onset, and so we put A, B, and C into
the database of known partially penetrant pathogenic mutations. Then comes along
not A/B and not B/C, but an A/C. So let
me ask the audience. Is A/C guaranteed to have
this late onset disease if A/B does and B/C does? What do you think? How many people say
yes, this person is expected to have the disease? How many people
say we don’t know? Yeah, so you know
this ABC problem, and the answer is we don’t know. So it’s a numbers game. You’re adding combinations of
partially penetrant mutations and the numbers matter. So not all pairs of
partially penetrant mutations will lead to
disease, and we only see a tiny fraction of the
combinations of partially penetrant mutations. So if we have 200 partially
penetrant mutations, there’s 200 squared
combinations. And we only see a few hundred. We’re missing lots of data
on all the combinations. And that is the
biggest challenge in next generation
sequencing for first tier newborn screening, in my
opinion– not the cost, not the time, not the
phasing, not the VUS, but the ABC problem. And time and time again, in
newborn screening programs, you see this ABC
problem haunting the use of next generation
sequencing for first tier newborn screening– lysosomal storage
diseases– family of serious genetic diseases,
many of which can be treated. One example is MLD–
you saw in the video. It’s caused by
mutations that lead to a dysfunctional
lysosomal enzyme. Without screening,
diagnostic odyssey is a problem, often too
late to start treatment. And expansion of newborn
screening programs worldwide to include a
subset of these diseases is a very big topic
of discussion, and in fact, is occurring
right now, thanks in part to the technology
that we developed here at University of Washington,
for better or worse. So just to tell you a little
bit about how these diseases are treated, you probably know about
enzyme replacement therapy, but I’ll go through it quickly. So all the proteins in
the cell are recycled. Why? Because I guess they poop out. They’re not that stable. Some last a few days. Some last a few hours. Some last a few weeks. They wear out. They have to be recycled. Typically, they’re broken
down and reassembled. So in the cytosol, you have
things like the proteasome that’s chewing up all the
proteins and recycling them, but what about the proteins
on the cell surface? They’re not in the cytosol. How do they get recycled? They get recycled by
enzymes in the lysosome. And not only proteins,
but glycolipids on the cell surface,
sphingolipids, proteoglycans. So it turns out
it’s a magic trick. So how do these proteins
on the outside of the cell go through the plasma
membrane and go through the lysosomal
membrane to come in contact with the enzymes
inside the lysosome? So they have to go
through two membranes, and they do that without
going through any membranes. It’s a magic trick. It’s like the ring trick,
where you take apart two rings. And the way it works
is it’s a magic trick. Not really. The proteins, you
get endocytosis. This vesicle seals off into a
vesicle where the proteins now on the outside of
the cell are now on the inside of the
lumen of the vesicle. And this guy, like
a soap bubble, fuses with this guy,
another soap bubble, so that you get one bubble,
and what’s inside each bubble becomes inside the same bubble. So it’s just a
fusion of memories. And now the enzymes that
were on the cell surface are in contact with the lumen
of the lysosome and the enzymes inside. So lysosomal storage,
there’s usually about 40 different
enzymes required for recycling of
all these goodies on the outside of the cell. If we have DNA
mutations that lead to dysfunction of
any one of them, we have a specific lysosomal
storage disease, often with different symptoms. So cells die for
different reasons because of the accumulation
of these metabolites. So enzyme replacement
therapy is pretty simple. We inject the recombinant
enzyme that we make in a factory into the blood of a patient. In the interstitial fluid,
it can come in contact with many cells of the
peripheral tissues. Not the CNS, because the enzyme
doesn’t go across the blood brain barrier, as you know. But it can just,
as a bystander, be taken up by endocytosis
in the fluid phase and delivered to the
lysosome from the outside. So we can just insert lysosomal
enzymes from the syringe needle into the blood, into
the interstitial fluid, and into the lysosome
of cells that are in contact topologically
with the interstitial fluid via the blood. So we just give you
the missing enzyme. And if we could get these
enzymes into the brain, we could do this
for brain diseases, but we don’t know how
to do that very well. People are working on Trojan
Horse type approaches, but nothing in the clinic yet. All right, enzyme
replacement therapy. You saw the movie maybe. You probably didn’t
see the movie about– actually, our
president mentioned John Crawley played by Brendan
Fraser at his first speech. I didn’t show you the
movie, because you know– He honored John Crawley. It was Rare Disease Day. You can watch the
movie on YouTube. John Crawley had two
kids with Pompe disease. He was a businessman. He quit his job. He found the scientist,
Harrison Ford. This was the first
and last movie directed by Harrison Ford. I think three people
saw this movie, and they all work for me. We went to the Varsity Theater. They said they were showing it. We got there, they said
they changed their mind. We had to drive out
to Auburn, Washington. We stopped at the
Muckleshoot Casino. And we got to Auburn, there
was a thousand people in line to see Avatar, and
we went to see this. And it was really funny,
because the business guy was yelling at the
scientist, like, how come you’re not
working fast enough. And I turned around and looked
at all my graduate students, and I said, yeah, see, that’s
what I’m talking about. So anyway, it’s a true
story about this guy that started a clinical trial
and got his kids treated, and they’re doing
pretty well, these kids. You can watch the YouTube. It’s kind of a remarkable story
about the treatment for Pompe disease. So what about if the
thing is in the brain, like MLD, the brain disease? So let me tell you just briefly
how this movie thing works. This is in PhD terms,
not in medical terms. They drain one pint of baby’s
blood from the placenta through the umbilical cord. This is called cord blood. So think of the
umbilical cord as Tygon tubing attached to this big
bag of blood that’s full of– you can’t take one pint
of blood out of a baby. Even a PhD knows that. And so they take a lot of
blood out of a placenta through the umbilical cord,
and it’s just baby’s blood. The fact that it’s cord
blood is just plumbing. Nothing special. They isolate
immature stem cells– they’re supposed to
be in the bone marrow, but they’re floating around
in the baby’s blood– by magnetic cell sorting. And they’re going to engineer
these cells to overproduce the missing enzyme. So they take an engineered
HIV virus, a lentivirus, just garden variety HIV
virus, and they engineer it so it’s not infectious. They insert a copy of the gene
for this enzyme with a very strong viral promoter,
CMV promoter if you care, into the genome, semi-randomly,
of these stem cells. Hopefully without
causing cancer. Then they destroy the baby’s
bone marrow with chemotherapy and radiation. And they do an autologous
bone marrow transplant. They give these engineered
stem cells back to the baby. They go into the bone marrow,
repopulate all the blood cells. So what does this have
to do with the brain? So obviously, you can treat
blood diseases this way, but it turns out the
immune cells, some of them have to get into the brain,
because you have macrophages, and dendritic
cells, and all kinds of things from your bone
marrow living in your brain. So they burrow through
the blood brain barrier of the capillaries,
and they reside in the brain. These lysosomal enzymes,
they can get into cells, but they can also
get out of the cell. Just reverse the process
I just showed you. So the idea is that
these cells in the brain, that came from the
bone marrow that are producing large
amounts of the enzyme through the gene therapy
to augment the process, are spewing into the brain
interstitium good enzyme, which are then taken up by cells
that did not come from the bone marrow by endocytosis. So this is called
cross-correction where one cell feeds the
other cell a lysosomal enzyme through the culture medium. So this cross-correction,
combined with overproduction of enzyme, is the basis for
the stem cell autologous bone marrow transplant with
gene therapy that you saw in the movie, developed
by Alessandro Aiuti’s team at the University of Milan– a very famous science
paper a few years ago. I think some of the
first applications of gene therapy in human health
will be lysosomal storage diseases. So we need newborn
screening for MLD. I’ll tell you how we
do that in a minute. All right, so now let’s
talk a little bit about how we do all this
screening, and it’s going to come down
to mass spectrometry. So you can do
fluorescent assays for– so here’s the deal. We don’t want to do DNA. We could measure the abundance
of the lysosomal enzyme by some sort of an ELISA,
because most mutations that cause disease lead to a non
stable misfolded protein that’s degraded, and therefore,
the protein abundance is usually low. So why don’t we just measure
the protein abundance? We could do that, but
we would miss cases where the mutation leads to a
dysfunctional enzyme and yet a stable protein
that’s not degraded. So we would have a
false negative problem. We don’t know how
bad that problem is. Maybe we would miss 5%
or 10% of the cases. Maybe that’s too much. But we decided 15
years ago to go after the function
of the enzyme, not the abundance
of the protein, and not the DNA sequence,
the activity of the enzyme. So now we want to measure a
bunch of enzyme activities at the same time in
a dried blood spot. And mass spectrometry is the
way forward for lots of reasons. You can do many enzymes
at once, as you’ll see. You don’t always have
a fluorometric assay for your enzyme of interest. Sometimes you do. So it would be nice to be
able to do this in general, not just for which you
can do fluorescent assays. And mass spec, I’m not
going to go into it as more precise than fluorescence. It gives less false positives. Mass spec is already used
in newborn screening lab, so all your amino acids are
measured by mass spec right now. That’s how they
detect tyrosinemia. That’s how they detect
PKU, et cetera, et cetera. So mass spec– and Ron Scott
got very excited about this when I mentioned this to
him about 15 years ago. So how do we do many
enzymes at once? It’s easy. We just take a
bunch of substrates, we put it in the blood
spot that has the enzyme. We have a bunch
of enzymes convert to a bunch of
products over time. And after a few hours, we
measure the amount of product. How do we measure the
amount of product? We count them in the
mass spectrometer. Just 1, 2, 3. We have a detector. So we have a machine
that separates molecules based on the mass. So we send all the
products with this mass, and it hits the detector,
and it goes ping, ping, ping. So we count the molecules. To determine the amount of
product, divide by the time is the enzyme activity. But we don’t know if the
machine is telling the truth. It gives us ping, ping,
ping, but maybe it missed a few pings. So we include an auditor. And it’s not a once
a month auditor. Every experiment
gets an auditor. So there’s no cheating. We’ll catch you if you cheat. So how do you
molecular auditing? It’s very simple. With mass spec, you add exactly
the same product chemical, but made heavier with
deuterium instead of hydrogen. So if you miss a few of these,
you miss a few of these, because they’re
the same chemical. It’s like if you’re counting
jelly beans in a jar, and you want to know
how many red ones. You put five red jelly
beans in the jar that are injected with
garlic, and so you’ve got 12 red jellybeans
in the jar, and then you taste them
afterwards and four of them taste like garlic,
but you put in five. So you know you missed 20%. That’s exactly what you can
do with an internal standard. Think about it that way. It’s an auditor. So it becomes
perfectly quantitative. In fact, the most quantitative
clinical technique I can think of. We use dried blood spots, so
we take a three millimeter punch of this dried blood spot. We mix it with a buffer with a
bunch of different substrates for a bunch of
different enzymes. We incubate to
allow the substrate to convert to product. We extract into organic solvents
just to get rid of the buffer salts, which interferes with
the mass spectrometry process. And then we inject this
into the mass spec. The mass spec is just a tube
that allows ions to go through with a certain mass. So we look for the
needle in the haystack just by selecting
on certain masses. Now, I can tell you
about how we know the thing we see is really
the product of interest, and I can show you all the
controls and everything, but there are ways to do that. For example, if you
leave out the substrate, and you don’t see the mass
spec signal for the product, you know the product signal
came from the substrate. It didn’t come from some
ghost peak in the blood. You leave out the substrate. You put in the blood. You don’t see the product. You need the substrate. It must be the product. That’s the idea. So here’s one of the substrates,
never mind all the details, it’s some enzyme
recognizable sugar. And then we make this product
where we can put heavy isotopes to make a chemically identical. So these are the same thing. This internal standard
just has deuterium. Without going into details,
the substrate– what we insist, of course, is that the substrate
detects the enzyme of interest. If we detect multiple different
enzymes with our substrate, it’s no good. So sometimes we need
to add inhibitors of off-target enzymes. We can do that, but
the assay is only as good as there
being a one to one correspondence
between the substrate and the enzyme of interest. I think you all know that. Just to show you the
spectrum of substrates, these are sphingolipids. If you can’t take
the phosphocholine off a single myelin
to make the ceramide, you have Niemann-Pick. You can’t take galactose
off, you have Krabbe. You can’t take glucose,
you have Gaucher. These are ceramides. How do we know the ceramides
came from these substrates? Because you have
ceramides in your blood. Because the R groups
on these ceramides are a little bit shorter
than the natural ceramides. And so we asterisk
these ceramides to come only from
these substrates via the enzyme of interest. So we’re sure that the
amount of this somewhat artificial ceramide came
from our enzyme of interest. Easy tricks. We have glyco
hydrolizing enzymes. We have sulfatases
that take sulfate. Anything you want, we
can do by mass spec. We have a kit, FDA approved
kit from PerkinElmer, to do six of these. [INAUDIBLE] mentioned
some states. About 30% of the
United States is now doing this live for newborn
screening for Pompe disease and MPS-I. Washington State’s
going to start next year. People at Children’s
Hospital are getting ready. Taiwan is doing this like crazy. Japan is starting, Australia. Europe not so fast. They’re a little bit
anti-screening over there. I don’t know. You can talk to them. Now, here’s the
ultimate solution. So I want to talk to
you about this MegaPlex. It was developed by
[INAUDIBLE] who’s sitting somewhere over there. And you’ll see her later. We can do 15 enzymes
and five biomarkers at the same time inside
two minutes by mass spec. So I think we could
do 200 enzymes, but I always tell Ron, why
don’t we screen for everything, and she tells me, because we
should only screen for things that we can treat. Otherwise, you
don’t want to know. And I agree with that. 10 years ago, I
didn’t know that. I wanted to screen
for everything, and she sort of set me straight. I’ll never forget
that conversation. But anyway, do we want to
screen for some genetic disease that we can’t treat? If the first kid dies, you
probably want to know why, so you can be on the
lookout for the second kid, but other than that, I
don’t know, probably not. Although it’s coming
through genomics someday. Anyway, we have a
choice, actually. We can look at the
enzyme activity, which I’ve been talking
about, which amounts to the amount of product
emitted as a function of time. Or the other way we can do this,
which makes a lot of sense, is if you’re missing an enzyme,
the substrate for that enzyme should accumulate. It’s not always that simple,
but for some diseases, the only choice is to
look for the accumulated substrate as a result
of deficient enzyme. So that becomes
a biomarker, just to measure the level
of the substrate. We do it all inside of 2.4
minutes by LC mass spec. We look at 15 enzyme products
with 15 internal standards, and we look at about
five biomarkers, also with internal standards. Everybody has their own auditor. There’s no cheating. There’s no lying. It’s the real deal. It’s not fake news. It’s real. We can do this
beautiful MultiPlex. And I won’t go into it. And we can add pretty
much lots of things that you want to this. We did a pilot study recently in
the Washington newborn screen– so you can’t just turn
this thing loose and go. You have to do a clinical trial. So if you expect MPS-II
disease, for example, you expect one in
100,000 patients. And if your screening
gives you 10 per week, you have a serious
false positive problem. You’re going to be
pulling your hair out sequencing all these DNAs,
scaring all these families. All these false alarms. A lot of sleepless nights. Big mess. The system can’t handle it. The families can’t handle it. So the false positive
rate has to be manageable. The best false positive
rates we have are typically, I don’t know, one in 10
screened positives turns out to be the real deal. Nine out of 10 turn
out to be false alarms. That’s really good. I’m talking about false
positives of one true positive in a thousand false
positives would be on this. So we’re always going to have to
do second tier tests to reduce the false positives,
but when it gets so bad, we can’t handle it. So we have to do a pilot study. So we ran a pilot– let
me show you some numbers. We ran a pilot study for five
different mucopolysaccharidoses in the Washington
State lab recently. We went to a hundred thousand
de-identified samples. So we can’t identify
these patients. We can’t offer them treatment. We just do a research study. We’re not allowed by law
to identify the patients. It’s kind of too bad, but
that’s IRBs and paperwork and all that stuff. It’s against state law. To do a prospective pilot
study, we need parent consent, and there’s lots
of places you can have a baby, so Ron
and I aren’t going to run around asking all these
parents for their consent. Although we thought about it,
but it’s just not practical. So we screened a
hundred thousand. We had 20 screened positives. That just means the enzyme
activity is below our cutoff. You can argue what
is our cutoff. When it’s low, it’s
a screened positive. The enzyme is low,
we’re worried. So we had 20 hits, and
we genotyped all these, and two have genotypes
consistent with MPS-II. We’re pretty sure these two
kids will develop MPS-II. It’s an X-linked disease. Two mutations are very severe,
well known in the database. So two out of 20. We had one MPS-IIIB
hit, genotype normal. Eight MPS IVs to probably
MPS-IVs, five MPS-VIs, all normal. This one is really remarkable. MPS-VII is thought
to be one in 300,000. So really rare. And we got one in
the first 100,000, and the kid showed up
at Children’s Hospital with MPS-VII. So you can see the numbers. We have about for
100,000, five diseases, we have about 30 hits. About four or five of
them are true cases. And that is a remarkably
successful pilot study, so based on this, we hope to
put some of these diseases forward for widespread
do more screening. And there’ll be more pilot
studies of this type coming in the next few years. In New York, we’re to be doing
a prospective pilot study where we can consent three large
hospitals in New York City and get 45,000 babies per
year with parent consent. So if only it were so simple. The enzyme’s high,
you’re healthy. The enzyme’s low,
you’re not healthy. It’s not that simple. It’s a continuum. I think you all know this. So you start off with
normal enzyme activity, you call it 100%,
population normal. As the activity gets
smaller and smaller, we generate a
continuum of events. For many enzymes, we
typically have more activity than we really need, and so
even if the activity drops to 20% to 30% of
normal, we’re still healthy for our lifetimes. Then we get into the so-called
pseudodeficiency regime where the activity drops
further, let’s say, to 10%, but you still don’t
get a disease. That would be a
pseudodeficiency, but you can see
it’s a continuum. What is normal and what
is pseudodeficiency, and what is low
and what is high? But officially, a
pseudodeficiency is a mutation, a mark
in the genome that causes a significant
reduction of enzyme activity, but not to the point that
it causes pathogenicity. But it is a broad
definition that has lots of interpretations,
so it’s kind of messy. So when the activity gets
below pseudodeficiency, we have diseases
that are late onset. We have diseases that start
as teenagers or adults, and then when it
gets even lower, we might have
infantile diseases. And when it’s even lower, we
might have a miscarriage rate. So it might not be viable. So it’s a continuum. So it’s not black and white. And we struggle with this. For some diseases,
like MLD, the video, the pseudodeficiency
is a huge problem. One in seven Europeans carry a
common mutation in their genome that reduces the activity
more than tenfold, so 1 in 7 squared, so 1 in 100 people will
read very low enzyme activity, and yet never develop MLD. The difference between these
people and somebody with MLD is MLD, you have less
than 1% activity. Homozygous for the
pseudodeficiency, you might have 5%. So we’re going to be
splitting hairs at the low end to try to figure this
out by enzyme activity. It’s not possible. So the only way forward is
to look at the accumulated substrate for MLD, which we
can do in this MultiPlex assay. The accumulated substrate
is the so-called sulfatide. This disease is
caused by a deficiency of an enzyme that
takes the sulfate off of the galactose on
the single lipid. This is a big
component of myelin. And these patients
lose their myelin. It’s a leukodystrophy
at a white matter loss. So we measure accumulation
of sulfatide substrate in dried blood spots by
LC mass spectrometry. I was very nervous, because
so sulfatides are maybe 5 to 10-fold elevated in urine. But we don’t collect urine
for newborn screening. I always wonder
what’s the big deal. But I’m told that collecting
urine for newborn screening is off the table in
the United States. They do it in Quebec– not the rest of Canada. Historical thing. I don’t know. But what’s the big deal
collecting urine spots from babies? I hear it’s a big deal, and
it’s not going to happen. So we then turned
to blood spots. And we found that sulfatides
were elevated in newborns with MLD, but not very much. So how do we know sulfatides are
elevated in newborns with MLD? So a kid is born. He’s healthy. He or she develops MLD when
they’re three years old. And how do we know what their
sulfatide levels are at birth? Because we don’t
know they have MLD until they’re three years old. Well, if we go to
Kentucky, that’s a problem, because they throw
away the blood spots at birth a few days after the kid’s born. I don’t know what’s going on. In California they
store them for 20 years. That’s a good
thing for research. So we can go back and retrieve
those newborn blood spots from the freezer of
California for those five kids over the last 10
years who developed MLD. And we can get a glimpse
of their sulfatide in a newborn blood spot. And that’s what you see in red. And this is a typical week of
data for 1,500 random newborns. And you can see that the
elevation of sulfatides at birth in the MLD is 0.3
versus sort of approaching 0.2. So this seems like a close call. Our best estimates
are that we’re going to get 1 0 to 2 or 3
false positives per week, which is probably too many. Be a hundred per year. So we’re scratching our heads
about what a second tier assay might look like
to stratify these hits. But we think screening for
MLD will be possible this way. But it’s going to be a
little bit challenging. This shows you what
we’re up against. Where we can’t do the enzyme
activity we have to do– let’s say a minor increase
in sulfatide levels in blood spots. All right. Now I want to turn to this
diagnosis/prognosis thing, which is going to– Children’s Hospital people
are going to be interested. So what happens often
in newborn screening is we get a kid with
low enzyme activity. How low is low? Well, we’re worried about it. We don’t know. Maybe the kid is not sick yet. When are they going to get sick? So they’re not sick yet. If and when they’re
going to get sick. That’s not diagnosis. I think that’s called prognosis. When are they going to get sick? I don’t know. When are they going to get sick? Maybe never. Maybe in five years. The parents kind
of want to know. The doctors kind
of want to know. The kid kind of wants to know. They were found by
newborn screening, which we didn’t do before. So that’s kind of a mark
against newborn screening– the fact that we’re putting
these families into this kind of horrific situation. It is not– I don’t know. You found out that
your kid might get MLD. Not so good. Now all the parents with
MLD kids, they think, let’s screen, because the only
chance you can save these kids is newborn screening. But all the kids families
that don’t have these kids, they say, well, wait a minute. I don’t know. They’re not sure. If I ask you, do
you want to know? I think you don’t know
whether you want to know. If I ask you in a week,
I still don’t think you know whether you want to know. Right? I don’t think you’ll ever
know if you want to know the answer to this question. I don’t know. You might know. I don’t know. Ask me. I don’t know. So when do you want to know? Do you want to know if
the answer is maybe? So there is a big problem. Right? I think you get the idea. And as we screen for
more and more things, it becomes more and
more controversial that you’re going to get
this when you’re that old. And maybe. But we don’t know. My hypothesis is that the
amount of enzyme is important. So the lower the enzyme, the
earlier you’re going to get it. I mean, that seems
to make sense. But people have sort of gone
against that hypothesis, because I think the assays
are not sensitive enough to see the difference between
1%, 2% kind of numbers. But with mass spec, you can
do it for the first time. So if you have at risk
for Krabbe disease from newborn screening. You go to David
Wenger’s lab where he measures the GALC enzymatic
activity in leukocytes. He cleans it up a little bit
by using leukocytes instead of a dried blood spot
so he can normalize for the amount of protein. But he uses a radiometric assay. Can he tell the difference
between, let’s say 1% and 2% versus 2% of residual– 1% or 2% of normal activity? OK. Let’s say population
mean is 100%. Can he tell a factor
of 2 between 1% and 2%? The answer is no, because
in his radiometric assay the difference would
be like five DPMs. No, he cannot. Statistically, significantly he
could not tell the difference between 1% and 2%. But with mass spec we can. Now how do I prove that? It’s easy. I take blood cells– immortalized leukocytes
from a patient with Krabbe disease that has no GALC
activity because they have– they’re missing the gene. They have large
deletions in both copies. And I mix in various
amounts, preconceived. I determine the amount
of normal leukocytes. So I determine the x-axis– the percent of normal activity. And you can see at 1% activity
I get this with 2 micrograms or protein or 5 micrograms. And at 2% I get that. And you can see the error
bars on triplicates. So I can tell the
difference between 1% and 2% unambiguously. In fact, I can tell the
difference even lower. To make a long
story short, we’ve looked at this
assay on a handful of screened positive patients. And we’re doing pretty
good with prognosis. We can tell you the kids
that are at highest risk to develop Krabbe disease. And so far they’re
developing symptoms in a few years of life. And we can tell the group
that is at less risk. And so far they haven’t
developed Krabbe disease out to 15 years. So we’re getting better from
measuring the residual activity very accurately in leukocytes to
follow up on newborn screening. We’re getting
better at prognosis. And we’re never
going to get perfect. Right? But we’re trying to
do better with very precise mass spectrometry
assays that you can’t do by other methods. Just to end in the last
five or six minutes– I know Ron is very interested
in lysosomal acid lipase. Let me tell you a
little story that’s kind of cool, kind of proud of it. At Children’s Hospital they do
a lot of assays for this enzyme. If you’re deficient
in this enzyme, you have cholesterol
ester storage disease, or also called Wolman’s disease. It’s a different type
of lipase deficiency. The standard assay in the
field is a fluorescent assay with a 4-MU palmitate. And the problem is there’s
many esterases in blood that can hydrolyze this
fleurometric substrate. So what they do is
they do two assays– one without and one
with, and an inhibitor that’s selective for the
lysosomal acid lipase. And the difference
between the two– the lysosomal acid lipase
inhibited, not inhibited– is the lysosomal
acid lipase component to the total serum esterase
activity on the substrate. So they do two
assays in parallel. But this is not good
for newborn screening because you’re going
to be subtracting two large numbers with errors. And you’re going to end
up with a small number with a big residual error. So I don’t think it’s going
to work for newborn screening. And I won’t bother you. You know how to
propagate errors. But anyways. So the first thing we
tried, we went back to the natural substrates
of this lipase, which are cholesterol esters
like this, thinking this would be a specific substrate. And we could measure
the activity– the hydrolysis of
this– by mass spec, because we can do
mass spec on anything. It doesn’t have
to be fluorescent. But it turns out
these substrates are hydrolyzed pretty slowly by– I wouldn’t say slowly. It’s just we cannot detect
the activity of these guys in a dried blood spot. There’s not enough lipase. The trick behind
this 4-MU palmitate is it’s a hand grenade. It’s a super reactive
ester that allows you to have very high
activity and see the lipase in a dried blood spot. At the same time it’s
very non-specific. So you see other lipases. But you have to use a
really active substrate to see the activity in a
small dried blood spot. So anyways, we made
a library of analogs of this fluorescent substrate. Or we made about 200 compounds. And we just said, look. If we can find one that’s
hydrolyzed well by the blood and is 100% inhibited by
the LAL-specific inhibitor– that is a LAL-specific
substrate. And we found one. And it works. So here is a bunch
of single assay. We don’t have to do this.
plus and minus inhibitor game. So children less than five
years old; 15, 12 adults; and six Wolman patients,
you can see the separation with a single assay
is really good. This can be done by mass
spec or fluorescence. Take your pick. So we found a specific
substrate for the lipase using kind of a library
approach of just getting lucky. Now just to end I want
to talk about something that’s really exciting. And Andy’s going to like this. So newborn screening
and proteomics. Proteomics is hard. You’re trying to
find the abundance of a low abundant protein in
a complex mixture like blood. So the ultimate needle
in the haystack problem. And so just– typically we
would do this by ELISAs– sandwich ELISAs. And there’s all kinds
of problems with that. You can talk to Andy about
thyroid thyroglobulin and auto antibodies and
interferences and problems in misdiagnosing
thyroid cancer with DELPHEA-type auto-antibody. [INAUDIBLE] antibody assays. You know, it’s hard to really
prove that your immunoassay is really selective. Right? The antibody could be
binding to other things. On the other hand, mass
spec is more selective, because you’re seeing it’s
got the right retention time on the chromatography column. It has the right parent mass. It has the right
fragment masses. But it’s harder to do,
and it takes more time. So we’ve got to fix that. The other thing is– well, let me just jump to
the case of Wilson’s disease. I’m very excited
about this project with [INAUDIBLE]
Seattle Children’s Research Institute
that we just started. So this is a disease
caused copper buildup due to mutations of the
ATP7B gene, which is a copper transporter in golgi membrane. So we have obviously no way to
do a functional– biochemical functional assay of
this metal transporter in a dried blood spot. Right? That’s off the table. The only thing I can think
of is to measure the protein abundance. It turns out as far as we
know, most Wilson patients have mutations that render
the protein unstable. And the abundance is low. Is that every Wilson patient? Probably not. But most of them. So we would miss a
few percent, perhaps– we don’t know exactly– of the Wilson patients that
produce stable abundant protein that is non-functional,
because we’re going to measure the
abundance of the protein. But getting 97% of
the Wilson patients is better than nothing. All right. So why don’t we do a
DELPHEA-type ELISA? Because it doesn’t work. You can try. We’ve tried it. It doesn’t work on low abundant
proteins in dried blood spots. These antigens are denatured. It just doesn’t work
when the protein is in very small amounts in
a very complicated mixture. So the only way is mass spec. But it’s not trivial,
because there’s many, many, many, many,
many, many different proteins in this blood sample. And we got this big needle
in a haystack problem. So what we can do
is the following. We can treat the dried
blood spot with trypsin to digest all the proteins
into the tryptic peptides. So proteomic techniques work
well on tryptic peptides. Not so well on whole proteins. So we chop the protein
into little peptides. And then we– the problem is
you’ve got 30,000 proteins. And now you have 3
million tryptic peptides. And you’re looking
for two of them. So you’ve got a big needle
in a haystack problem. If you put it into
the mass spec, you can do everything
you can think of. You’ll never find those peptides
for low abundant proteins, unless you’re incredibly lucky. So we have to enrich. So we use monoclonal antibodies
that bind to the two target peptides from the
Wilson protein. And we pull those peptides
down with magnetic beads. And even then it’s still
a mixture of peptides, because the antibody
enrichment is far from perfect. And we’ve got a
gazillion peptides. So then we put it
into an LC mass spec. And we look for the signature
peptide from the Wilson protein that has the right retention
time on chromatography. It has the right parent mass. And it has the
right fragment mass. And that is, in fact,
the Wilson peptide from the Wilson protein. And we do two such
peptides as a double check. And it works. The problem is is it takes
about 20 minutes per baby. And so the goal of
the next few years between me and [INAUDIBLE]’s’s
lab is to get this down to two minutes per baby. But how are we going to do that? To make a long story
short, we have some tricks up our sleeves. We’re going to use tagging
so that every baby gets a different tag. So then we can do– if we do 10
babies at once in 10 minutes, and we tag them differently
with mass signatures, we do one baby per minute– 10 at a time. All right. We just can’t mix up the tags. And then we get the wrong baby. But that’s always possible
on newborn screening. So we’re going to
use tagging strategy, and then put a bunch
of babies together into the same 10-minute run. And then that gets you down
to one minute per baby. And that’s the goal of
the next five years. Summary. I think I’ll skip that. Skip that. Just to thank a lot
of people involved. So the story started
about 15 years ago when my wife was pregnant,
second child, amniocentesis. And I asked a lot of questions
like what are you doing? And they said something
like spinal bifida and Down’s Syndrome. And I saw that
movie, Lorenzo’s Oil, and I thought, what
about all these diseases? And I’m an enzymologist. So why don’t we screen
for everything at once and just get it over with? And then Rona told me, no. It’s not that simple. And then I met Ron. And Ron said, this
is a good idea. Let’s have a go at it. And we should work on
lysosomal storage diseases, because this is the next
frontier of expansion of newborn screening with
these new treatments. This is where the
field is going. So that was– Ron
was good for that. He whittled it down to what
was needed in the field. And here we are 15 years
later with an FDA-approved kit for doing this. 30% of the United
States is live for this. The rest will come in
the next five years. All of Taiwan, Australia,
Japan is starting. Let’s see what happens. A lot of people
involved in this work. I want to call out a few people. Sophia Masi developed
the lipase assay. She just got her MD PhD
in our program here. Joyce Liao is from Taiwan. She’s sitting back there. She’s now in the clinical
chemistry fellowship program with Andy– one of the best
graduate students I’ve ever seen in my 35 years here. She’s the one that– and she spearheaded this
Taiwanese lysosomal newborn screening program, which
is the most progressive newborn screening
program in the world. So the only graduate
student better than Joyce is Shin-Ying, who’s
sitting back there. She’s superstar. And she’s like, [INAUDIBLE]
superstar, let’s say. These are two of the
best graduates students I think I’ve ever had in
40 years I’ve been here. Shin-Ying has
developed the megaplex and a bunch of other things. There’s a lot of
other people involved. I also want to call out these. These are phenomenal
graduate students. I think Shin-Ying wants to
go into a fellowship clinical chemistry program also. We’re talking about that today. Here’s some people. I’ll just end–
meeting the families is kind of interesting. What these families
endure with these diseases is just unbelievable. And they go around
thanking us for all of the wonderful research we do. And I just wonder– I don’t know. My life is good. My kids are healthy. And yet they go around
thanking us for all the– and we get their thanks, too. And I just want to give
all my thanks back to them. And these are my heroes. Right? What these guys go through
is unbearable to me. And we have to do better. So newborn screening, I think,
and gene therapy and things like that is the
only way forward. And we can talk about how
controversial this is. Or we could do nothing. And I don’t think we’re
going to do nothing. So we march through
this controversy, and we see how we do. OK. Thank you very much. [APPLAUSE] INTERVIEWER: Thanks, Mike. It was great. I don’t know if you can hear me. [INAUDIBLE] MICHAEL GELB: Yeah. INTERVIEWER: So you have
an FDA-approved kit. And you want to add the
next couple of diseases. What do you got to do? You got to go all the
way back to square one? MICHAEL GELB: I guess so. But I would say yes and no. 95% of newborn screening
labs are OK with LDTs. So there’s a couple of states
that require an FDA kit. They’ll be behind. And so this and that. Yeah. It’s kind of a
problem, but not really in the newborn screening world. They’re used to LDTs. As you know, if the FDA shuts
down LDTs, people like you and newborn screening
labs have major problems, because we have hundreds and
thousands of these rare disease assays for which no
FDA kit is in sight because it costs $3 million
to develop a kit that’s going to be used 300 times a year. Right? It’s just not possible. So let’s hope we
don’t shut down LDTs. And in newborn screening,
LDTs are a big thing. More than you think. So I think it’ll be OK. But if the market is
big enough, Perkin Elmer and other companies
will spend the money for an FDA-approved kit. That’s what happened for LSTs. Took a while. They didn’t want to do it. I think they were– anyways. Yeah. They finally did it. They didn’t want
to spend the money. Till the market– The market, until
you spend the money. It’s like a catch-22. Eventually they spent the money. Big problem. But labs will do LDTs. LDTs is the way out. Let’s hope for LDTs. You all know about LDTs. They’re a vital
thing in what we do. And it’s impossible
without them. [MUSIC PLAYING]

Leave a Reply

Your email address will not be published. Required fields are marked *