Rare Events with Large-Impact: Bioengineering & Clinical Applications of... (Mehmet Toner)
Sep 5, 2023
"Rare Events with Large-Impact: Bioengineering & Clinical Applications of Circulating Tumor Cells"
Mehmet Toner
05/08/15
Center for Nanoscale Science and Technology (CNST)
Thank you so much, and it's a wonderful introduction.
Maybe I should quit.
And thank you for being here.
I think I was here a few years ago,
and it's always great to be here and see the progress,
tremendous progress, over the last few years.
You guys are a bioengineering department in steroids.
I mean, you guys are growing faster than any place I've
known, and in a very nice, balanced way
and big breakthrough.
So it's a pleasure to be here with colleagues and friends.
It was about 1984 or 85, I was at MIT student.
And my advisor, I was doing low temperature biology.
That's still something I do.
And my advisor's advisor, my advisor
was from UC Berkeley, Ernie Cravalho.
And his advisor, so my academic grandfather, came to MIT.
And so I got, at the time he was the chancellor of UC
Berkeley, very famous heat transfer thermodynamics
professor, who's really considered
to be one of the first who got into nanoscale processes
and events in heat transfer and thermodynamics.
He was very famous, his scientific work
was very well-known, Professor Tian.
So I get to have a lunch with him.
And one of the things he told me is,
I'm trying to emulate as much as I can.
He said, Mehmet, go to extreme, you
will discover interesting things.
So that really stuck with me over the next 30 years or so.
And many of the time, we try to go to extreme.
We don't always succeed, but I love
going to minus 196 degrees Celsius with living cells
and bring them back.
And we were last night talking about,
I want to build a colony in the solar lunar south pole
for all mammalian species germ lines,
because it's at cryogenic temperatures
for the last few billion years.
It doesn't see the solar.
So it's going to that extreme.
And part of that, about 10 years ago,
we were doing microfluidics.
And I was told, no, microfluidics,
the fame comes from processing small amounts.
You can't do big amounts.
It's whoever does smaller amount is what's important.
And I was in a hospital, and I want
to get into more applications in blood and diagnostics.
So we said, can't we use microfluidics in an extreme way
where I call it putting a garden hose through the chip,
a microfluidic chip, where you have
to process whole blood, 10, 20.
In fact, we have technologies now.
We could even do liters of bodily fluids
through these channels.
And so we started looking, what would
be an exciting application?
And as I started actually reading the literature,
circulating tumor cells came up a lot.
And not in my context of microfluidics,
but people have been working on circulating tumor cells.
And they were using some of the old technologies
to find them, not so successfully.
And I thought this might be a great opportunity,
and again, an extreme science in the sense
that these cells are one in a 10 billion blood cells.
So they're extremely rare.
And so we started thinking about it, and going to Blackboard,
and designing ideas.
Those are the best, most fun part of the research,
the early phases of when you really
don't know what you don't know, and you're excited about it.
And then you realize it's pretty difficult.
And more recently, we got into clusters.
So it's even rarer than rare cell.
And so what I want to do today is just
give you a sense of the technologies we developed.
And some biological applications,
not a lot of the clinical stuff today,
but some of the biological applications.
Like anything, nothing we think is actually original,
or nothing I think is original.
What Rashid thinks is always original.
And this goes back a few millennia.
And Hippogritus was, and later, Dr. Galen,
very famous Roman doctor.
And of course, you all know Hippogritus.
They defined all diseases in four fluids,
and they called them humors.
There is the fowl blood, yellow bile.
Those are, you know, make sense.
And then there's the black bile.
Black, in Greek, is melan, and bile is colicus.
So melancholicus, and it was given for two diseases,
cancer and mental diseases.
And that's why that guy looks, of all of those,
it's the most unhappy guy there.
And the reason why they called it black bile,
at the time, cancer was diagnosed very late,
and typically breast cancer.
And it will be, it's called fungating, or wounding cancer,
because it will break out of the tissue into the skin,
and it will look black.
So they thought there was a black fluid circulating,
causing cancer.
But of course, Romans didn't do autopsies,
so they couldn't really do it.
They did some animal experiments.
They couldn't figure it out.
But then comes Vesalius,
very famous father of modern anatomy, in 1500s.
And he started doing the first extensive autopsies.
And he was coming from the school of Dr. Galen,
and believed in the four humors of disease processes.
So he's done a lot of these autopsies,
but he was never able to find the black bile.
And so he was very disappointed.
He published his results, and then life moved on,
and a few more hundred centuries later,
a resident at the University of Monash, in Australia,
has the observation, post-mortem,
a 38-year-old male cancer patient.
And you can see Dr.
You can see Dr. Monash sitting there in the laboratory.
He was a resident at the Monash University,
and he published a wonderful one-page article in 1869.
And this sentence is from that.
The fact of cells identical with those of the cancer itself
being seen in the blood may tend to throw some light
upon the mode of origin of multiple tumors
existing in the same person.
This is pretty visionary for the time.
And today we know that seven million people
die from cancer every year.
We know that one in two men and one in three women
in their lifetime will be diagnosed in this room
with cancer, so the numbers are pretty nasty.
And we know that nine out of 10 deaths in cancer
is because cancer spreads.
Primary tumor very rarely kills.
So we know it goes, we know there's a black pile.
And he showed here in the pathologist's images,
this is the primary tumor.
You can see the nucleus, you can see nuclei.
This is the peripheral blood sample.
You can see three cells that look exactly
like the tumor cells,
and those are the circulating tumor cells.
Since following almost 150 years,
we learned a lot about circulating tumor cells.
And what happens is primary tumor forms.
As soon as it's the size of a few hundred micron,
it's vascularized, and then these cells
start going into homeostasis with the body.
They are sloughed off from the primary tumor,
and they go, and once in a while, it's like a zip car.
Once in a while, they can find the right niche,
nice location, and they don't die.
They're home to that location, and they grow.
And they colonize.
So this process now is well-known.
It has 20-some steps, and that are very critical
in the introvasation and extrovasation of these cells.
And imagine, if you could find these cells in circulation
from a blood test, which is minimally invasive.
You do it every year, multiple times
when you go to your doctor.
You can imagine the ways you can personalize
the treatment of cancer.
You can diagnose it, you can monitor it,
you can look at the changes,
whether or not somebody's responding.
So it would be a very powerful tool,
the bioassay, that could change the way
we deal with cancer.
Almost like what we do in infectious disease,
you could turn it into what we call a chronic disease,
because you can monitor it very closely.
And now in cancers, we have so many different drugs,
these decisions are far more important than 20 years ago.
Time is of essence.
And so, the presence of these cells are known,
the biology of it we know, but we can't find them.
And so let's look at some numbers here.
If you take a tube of blood, it is 10 milliliters of blood.
You have five liters of blood in your body,
that's 10 milliliter.
There are about 50 to 80 billion cells in there.
And there are one, two, typically,
if the technologies I show you,
we can find one tumor cell in the tube of blood.
So it's a sensitivity of almost one in 80 billion.
And so if you were to look at each cell for one second,
blood cell, blood cell, tumor cell,
do you know how long it will take you
to look at the tube of blood?
Any guesses?
Days, weeks, years?
Years?
Years.
Years, sold, gone, any other?
3,470 years.
And if you take one second, the technologies,
the last one I will show you is interrogated
at about 50 million cells per second.
And so it has to be fast.
If I tell you, in the entire world,
there are about seven billion people.
That's one milliliter of blood, one-tenth of the tube.
If I tell you there is one guy with three ears
or three eyes and go and find it,
that's a pretty challenging task.
And so it is, you know, I'm just trying to get you
to understand that it's rare cells that are very important.
So people said that they are not in the circulation,
immune system removes them, complement removes them,
it's not useful, maybe you find it 10% of the time
in metastatic disease, then it's clinically not exciting.
So, and people used mostly, I won't go into details of it,
technologies that are based on size, density,
biophysical parameters, differences, bulk differences,
but tumor cells vary.
Or they took magnetic sorting techniques
or flow cytometry, tried to apply to a rare cell.
Technologies developed for common cells
and tried to apply it to rare cells.
What, the way I look at it, or I looked at it 10 years ago,
I said, look, these guys are trying to race
in Formula One with a Hyundai.
You can't win that race.
You need to build a Ferrari equivalent of it.
And then we built the first CTC chip,
circulating tumor cell chip.
And that was in 2007.
And it made it into this review article by Pantel.
And so I'll go through some of the design principles quickly
because I want to get to the third generation chip.
That's what, and the first generation,
this was a micropost.
It was about the size of a credit card.
It was etched using deep reactive ion etching and silicone.
And you see the post.
Posts are about 40, 50 micron diameter,
100, 120 micron tall, very nicely manufactured.
And we covered the post with an antibody.
85% of all neoplastic diseases, cancer,
is in epithelial cells in different organs.
So we covered this with an epithelial antibody, EpCAM,
epithelial cell adhesion molecule.
All epithelial cells have these antigens on their surface.
We put an antibody.
It's like a very precise immunoaffinity chromatography.
There are hundreds of thousands of these posts.
And the flow rate here is about two milliliter per hour.
So it takes about four or five hours
to process a tube of blood.
It interrogates about 1 million cells per second.
Each post, every second, interrogates only 10 to 20
cells.
So you make them by 100,000 posts.
I won't go into fluid mechanics.
They're distributed in a clever way.
And you break the streamlines.
And the whole blood goes through this, antibody coated.
And here is the first lung tumor cell
we captured in our first patient experiment.
In the next slide, I'll show you the inner working
of the device in a video, if it works.
Half of the time, it doesn't.
And whole blood will go from your left to right.
You won't see the whole blood because we didn't label it.
You have to take my word.
We spiked in lots of tumor cells in it.
It's not rare anymore because we don't want to wait three hours
here to see a cell so that you see the flow of conditions
and how it captures.
You see how it goes around there.
And once in a while, they stick.
There are five.
That one is stuck there.
I think this one gets caught right there.
Or yep, that one.
And actually, they stick all around it
because it's laminar flow.
And there's some flow separation in the back.
But it's not that in these flow rates.
It's not that critical.
Sheer stress is very low in these devices.
We don't push it very hard.
And that's a very straightforward idea.
What this did is we did a targeted therapy study
published in the New England Journal of Medicine.
We showed that we can get these cells very routinely
in metastatic cancer patients in lung, prostate, and breast.
So it kind of changed the dogma, the belief
that they didn't exist in circulation in enough numbers.
And that started a whole group of investigators
developing technologies.
The problem with this technology was expensive
to manufacture at the time.
So we wanted to see the blood because surgeons
love to see the blood as it goes through.
And so we designed a second generation,
which is based on herringbone structure.
In this case, we don't have posts.
But what we have is a flat bottom.
And the top surface has these herringbone structures
originally designed for molecular mixing by white sites.
Basically, you get this chaotic advection mixing patterns.
But this could be manufactured in plastic.
And the Hollywood picture, the embarrassing picture
Rashid showed, came from our funding
from Stand Up to Cancer.
It's a Hollywood organization that's
a philanthropic organization that funded our team
as a dream team to do a large-scale clinical trial.
We just finished that.
I won't go into details of that.
But we could manufacture this chip now
in plastic controlled in reasonable amounts.
We could put it in the hands of other people to test it.
And in the next video, you will see the inner working of that.
It will be a little bit different
because the cells will be tumbling.
So they'll be in and out of.
You see how they come and they stick to top and bottom.
That's, again, antibody coated.
It's positive selection.
So as we have gotten into this, we
realize the tumor is smarter than we think.
And this is a gallery of images from a number of patients.
First of all, look at this.
This is the same lung cancer patient.
Green cells here are cytokeratin.
So those are tumor cells.
Blue is the nucleus.
CD45 is common leukocyte antigen.
So that's a white blood cell.
Tiny.
Look at this.
That is, like, mega size.
But in the same patient, look at this tumor cell.
It's smaller than leukocyte.
And morphology is changing.
It's same in prostate cancer.
Then we looked at the phenotype of these cells.
Some of them were dividing, proliferating.
Some of them, we even got it as it's
dividing into two cells in circulation.
Look at it.
It's in telophase.
Some of them are in mitosis.
Some of them, many of them, are going through apoptosis.
So they're committing suicide.
Because these are epithelial cells,
they are used to sticking to surface.
Once you take them off the surface, they don't like it.
They commit suicide.
It's called aneoicus.
And there's always a Greek word for every biological process.
And apoptosis is committing suicide.
Some of them are covered with platelet.
So they are stealth.
Some of them go through what's called EMT, epithelial
to mesenchymal transition.
I'll talk about it, because it's a very important
fundamental issue in cancer.
Some of them come in clusters.
I'll talk about that, too.
So what we realized that they come like Ben and Jerry's
ice cream, in multiple different flavors,
and positive selection of going just after EpCAM.
And you need to express it high level.
Low EpCAM, you miss it.
No EpCAM, such as an epithelial melanoma, no marker.
How do you find melanoma cells?
There's no marker for them.
If they're covered with platelet, you don't get it.
So we had to come up with an idea.
Here, we are looking in that needle in a haystack.
And with positive selection based
on antibody against tumor surface characteristics,
you're trying to go and find the needle from the haystack.
So we said, can we turn this around?
Can we find all CTCs without even knowing
if the person has cancer or what type of cancer?
And so independent of their size and phenotype.
Of course, if you can go after the haystack,
you could do that, right?
If you go and throw the haystack in garbage,
you're left with tumor cells.
But your problem now, in a two tube of blood,
that's what we process, about 200 billion cells.
You need to throw away 199,999,999,999 cells
without losing that 1 to 10 cells.
As you know, when the fluid moves, of course,
that rare cell is going to go to garbage
with the rest of the haystack.
So how do we get around it?
What would you do?
At the time, very bright.
Now, as a faculty member at UCLA postdoc,
I think you overlapped with him, Dino DiCarlo.
And we were doing some experiments.
And the idea was we were pushing fluid
in microchannels a little bit faster than the average guy.
Because it's always laminar flow,
Reynolds number of less than 1 in these microchannels.
We were going to 5, 10, 20 Reynolds numbers.
Still laminar flow.
So you see one channel here.
You see the cells.
And we're going to increase the flow rate.
You see, if they come in the streamline here,
they, of course, go in that streamline.
They don't switch streamline.
That's laminar flow definition.
Otherwise, the physics will be totally out of the window.
Look at it.
We increase the flow rates.
Now they started actually being in only two lines.
They shift.
And they find two locations that are equilibrium locations.
I won't go into details of it.
But what happens is when the particle size
is the same order of magnitude as the channel,
the wall effects, as well as the shear gradient
that the particle sees, pushes towards the wall.
The wall pushes it back.
And they find these beautiful equilibrium positions.
And so we looked at it and said, oh my god.
Particles are behaving independent of each other.
If I want to throw billions of cells in garbage
without affecting the, if this is a tumor cell that's
a white blood cell, I want to throw this in garbage.
I don't want this to go with that.
This is the old particles, and it's high throughput.
We can put billions of cells through these channels
within minutes.
And so we said, OK, can we turn this
into a technology that could find rare cells?
So what we said, red blood cells, platelets are small.
We could make a mechanical filter for those.
White blood cells are in the same size as tumor cells.
So we added microbeads to blood, tiny magnetic beads,
coated with antibodies against white blood cells.
Those reagents are very well known.
They don't change.
So it's not like the tumor thing, they don't change.
We know the reagents.
We know the antigens.
And so we mixed the blood with microbeads
and a bunch of different antibodies.
And the first part, it has a three-stage process.
The first part is a mechanical filter.
It's the pachinko device, like a Japanese game.
You drop the money, small money goes straight.
Bigger money bangs into walls and goes to the left.
And originally described by Dr. Austin
over at Bob Austin at Princeton,
we scaled it up to whole blood and so on and so forth.
It's called deterministic lateral displacement.
So whole blood comes, buffer comes here.
Red blood cells, platelets, and free magnetic beads
come out to garbage.
All the nucleated cells, white blood cells
and tumor cells are deflected in the pachinko.
Then they come to the inertial focusing device.
By playing with the geometry of that,
I won't go into details,
we can focus to wherever we want to.
And so we lined them up.
And then they come to the magnetophoresis system
because now they are independent of each other.
You can magnetically deflect these cells
because it's so well aligned,
all you need is to deflect them five micron,
you separate it.
It's an extremely high throughput, extremely sensitive,
and every particle behave independent of each other.
It's easy because you pay $500,
your artist makes this beautiful figure for you.
And it really looks nice and it makes cover pages
and good on your walls.
But you need to make it.
So this is the final chip.
It went through three years of evolution.
Parts of it was, each part made in different materials
and different ways and then we combined it.
I won't go into the history of that.
And this is done on a CD because Sony makes it.
We didn't want to change their form factor,
it doesn't spin.
And this is, it's symmetrical.
This chip processes about 20 mil blood in an hour.
And whole blood.
And it sorts about 50 million cells per second.
Whole blood comes here.
That's the pachinko.
There are 202 million microfutures in there.
This is the world's most complicated plastic manufacturer.
The cost now is probably $20, $30.
About two million microfutures.
There are 256 pachinko devices there,
micro devices parallel.
And another 256 here.
Red blood cells go to garbage.
And we have all these resistances to adjust the flows
and so on and so forth.
This is the first inertial focusing area.
This is the most critical part of the device.
Then it comes into the magnetic region.
There are magnets under this.
And I'll show you later why we inertially focus it again
and then magnetically deflect twice.
And we collect the rare CTCs here.
Red blood cells, junk goes out of here.
White blood cell, junk goes out of there.
What's the third certainty region, the smallest?
Which one?
This one?
Yeah.
That's just the resistances to adjust the flow.
So that as we put the complexity onto the chip,
the peripheral complexity became very simple.
So the processor didn't have all the pressure controls
that required.
So it evolved very fast into a chip that has almost,
we have actually, the new versions have even more
complexity, but essentially this is the way it works.
I won't go into too much of the details of this,
but just to give you, we put green and red label,
white blood cells, red blood cells,
and then to see if we can show you,
this is one of the 256 Pachinko devices.
They come together, red and green.
As you go downstream, that's waste.
Red cells are separated from white blood cells.
White blood cells go into inertial focusing,
randomly distributed in the channel.
Then it's a single streak image.
These are streak images of the fluorescent images
into a single file.
And this is the magnetic deflection.
Single file, we make it two to get the high throughput
faster, and then you can see in the magnetic region,
white blood cells collected, CTCs go to collection.
And this is a video of the, again this is,
we spike many cells so that we can see
during the time period I'm giving the talk.
And if you are crazy, come over to the lab,
wait for two days under the microscope,
you can see one of these.
And these are white blood cells coated with beads.
This is the earlier version.
The problem we had, the processing magnetic field
was so sensitive that even if we had one bead on a cell,
we would move it to the wall and it would clog.
That's why we did the version I showed you
that has two magnetic circuits.
The cells come in from here, stabilize here,
inertially focus here.
The first one is a low magnetic field.
So we separate cells that have seven to eight
to 10 beads on their surface.
They move about 50 millimeter per second
and we have a separation about 10,000 cells per second.
And this is after RBC platelet removal.
And then here is the second part of the inertial focusing.
This is the high magnetic field.
In that we can, our sensitivity is actually half a bead,
we can remove that cell.
So it's a very sensitive.
So that's where the rays, they go around twice.
And then you worry about voltage chip coupling.
If you're doing things in the hospital,
and your colleagues are not engineers,
they wanna use this.
If you don't give them something usable, they yell at you.
And so with Johnson & Johnson,
we are manufacturing this to a point
where it can be scaled and we are running
lots of clinical studies.
J&J is doing five center clinical trials in this year.
So this black ugly plastic is just to put
all the lines in and out.
And there's a processor, the chip sits right here.
Blood is here, buffer is here, that's waste,
red blood cell, waste, white blood cells.
And then here in the middle,
you collect the circulating tumor cells.
They come with some other contaminating cells.
There are other progenitor cells,
fibrocytes and some white blood cells.
Typically in one mil of blood,
we have 200 to 400 cells within which there are tumor cells.
I love this picture because this is a melanoma,
so you couldn't capture these cells with positive selection.
And it's almost 250 years after
Ashworth's original study in 1869,
where he showed the primary tumor.
This is finally the last bit.
That's the primary tumor, that's a melanocyte right there.
You see the nucleus, you see the nuclei,
and that's in the peripheral blood.
And so that is, for first time,
this is clinical level diagnostic quality cytology.
You couldn't do this with the other techniques
because here, cells are in a tube.
We could culture them, we just published a paper.
We can do standard cytology,
we could do single cell sequencing.
Because they are isolated, pristine, live cells in a tube,
we could do anything we want to do,
and we have done everything that you can imagine, of course.
But what I want to do in the remaining time
is to talk more about the mechanism of metastasis.
But first, I want to show you that the technology,
as I promised at the beginning,
separates these cells independent of size and EPCAM,
or surface antigen, tumor antigen.
We have done this with thousands of cells now,
with real patients, these are not cell lines.
We measured the EPCAM expression on the surface of CTCs,
and as we defined by a number of markers and cell diameter,
you see that these are white blood cells,
many of them are in the same range.
So size separation is you're gonna lose most of your cells.
If you look at EPCAM expression,
this is typically, this is the region
where you could capture with positive selection.
So all these cells on this side, you will be losing.
And so it shows that we can capture
a very general phenotype and size of these cells.
And then we have done mass flow cytometry of these cells.
This is just one example from prostate cancer,
to look at 50 different protein markers on these cells,
and show that many cells that don't have EPCAM
actually has PSA, PSMA, that's marker for prostate.
So again, you wouldn't be able to get these cells.
Many of the cells that have EPCAM
don't have the other markers,
so it is very heterogeneous population of cells.
And so this is the general technology.
So what I would like to do next is talk a little bit
how we can use tools like this
to understand biology of metastasis.
One part is clinical use.
The other, maybe more exciting part,
is you can use these tools
to understand the biology of these cells.
There are two overall hypotheses for metastatic cascade.
One is put forth more recently by Bob Weinberg at MIT
called epithelial-mesenchymal transition.
Epithelial cells go to mesenchymal phenotype.
They become more motile, more aggressive.
They slough off from the primary tumor,
circulate, and cause metastatic process.
The other one, which is much earlier,
by Fidler in 1950s and 60s,
they called it tumor emboli, embolus,
and they believed that it was clusters of cells
that slough off from the primary tumor, not single cells.
And they get clogged in capillaries
because they're fat, big,
and they grow into the capillary.
And so that was a much more mechanical theory.
So we wanted to go after the mechanism of the process.
Now we have a tool to do that.
The EMT has been shown in animal models,
but it's never been shown in human cancer patients.
So we have done a breast cancer study.
We have developed epithelial markers,
seven markers in green,
and mesenchymal markers in red,
three different markers mixed.
And this is RNA-ish, so we are staining for RNA.
There are no good antibodies for this,
and it took a long time to optimize
all this process, of course.
Green is epithelial, red is mesenchymal.
This is the same breast cancer patient.
Look at it, some cells are all green,
some cells are all red, some are in between.
So this is the first demonstration
of epithelial mesenchymal transition
in an actual patient.
Here is an index patient.
That's so amazing.
And look at the CTC numbers
as a function of time during treatment,
PI3K treatment at the beginning.
Blue is mesenchymal, red is epithelial phenotype.
Initially, lots of CTCs in circulation.
Almost all of them are mesenchymal phenotype.
Responds to treatment beautifully.
The disease disappears, evaporates.
CTCs become almost all epithelial.
They dynamically shift their phenotype.
Stops responding.
You see in real time, you could see
that they're not responding and you change treatment.
Responds beautifully again.
Suddenly, as it stops responding,
it becomes all mesenchymal.
Responds to treatment, it becomes all epithelial.
And then the patient expired, it was all mesenchymal.
So you will call, Bob Weinberger was right,
EMT is driving the cascade based on this.
Are those days, are they extensive?
Months.
The number of CTCs is decreasing.
It decreases.
And they're changing phenotype.
They become more, they are responding.
They become more epithelial.
They stop responding.
The tumor becomes much more mesenchymal.
Responds again, becomes much more epithelial.
So clearly, epithelial mesenchymal transition,
as Bob Weinberg said, is driving the process.
And we have multiple patients like this.
What was very interesting is,
what was the cancer in the previous?
Breast cancer.
And when we looked at these cells,
when we saw a lot of mesenchymal phenotype,
we saw a lot of clusters.
But this chip was not designed to capture clusters.
There's mechanical filter, all kinds of stuff.
But we saw more clusters.
And then we almost saw no clusters
when it was all epithelial phenotype.
So it doesn't differentiate.
From an engineering perspective,
they're totally coupled right now.
Because first of all, when it is mesenchymal,
the belief is that the cells become epithelial,
attached to each other, become mesenchymal,
it becomes single and move out as single cells, motile.
And it's just the opposite of that.
So cluster also correlates.
More clusters, not good.
Less clusters, it's better.
Then we looked at patients.
And see if there's a correlation between survival
and the number of clusters.
And so we could capture these cells.
And if the patient had single CTCs,
survival was much better.
The red one, if this patient
in three longitudinal time point had clusters,
survival was terrible.
So clearly, clinically, breast cancer, it correlated.
In prostate cancer, it was even more remarkable.
These patients didn't have clusters in their circulation.
These patients had clusters in their circulation.
So it was both EMT operating and clusters.
Then Nicola, one of the postdocs,
developed this wonderful animal model.
So we wanted to know,
if clusters are coming out of the tumor as cluster
or as single cell and then grow in the circulation.
Because some of them are proliferating.
And that is fundamentally very important
to our understanding.
So what he did, he took an animal model.
He put breast cancer cells.
He labeled half of them red, mCherry,
half of them green, GFB.
Same cell, no difference, just different colors.
Mixed one to one.
Put it in the animal.
And this tumor goes to lung very efficiently.
And as you can see, if the tumor metastatic sites
or the ones in circulation are two color,
that means it's coming sloughing off as cluster.
If they are single color,
that means they are sloughing off as single cells
and then growing.
We have done many other controls,
but this is just a key experiment.
When we look at the circulation,
first of all, 97.5% of all cells were single cells.
Only 2.5% of cells were in clusters.
But almost all clusters were dual color.
So clearly they were coming sloughing off as clusters.
Then we looked at the efficiency of metastatic cascade.
Single CTCs, when we look at the metastatic loci
in the lung, we counted how many are single colors,
how many are multi-color.
If they are coming from a single CTC,
it should be single color.
If they are coming from a cluster of multiple color,
it should be multiple color.
And we counted, although 97% of all CTCs
were in single cell, 50% of them roughly of the lung
were coming from single cells.
2.5% were clusters, but 50% of the metastatic sites
were coming from clusters.
So if you do the simple math,
clusters are about 50-fold more metastatic than single cells
and experiment like this, animal model,
and a tool like this, you can go after
this kind of very fundamental understanding.
So that's quite exciting.
Then we said, what is in the clusters that is causing this?
So we did the RNA sequencing of single CTC clusters.
Is there a specific difference
that's upregulated in clusters?
And so we took the single cell CTC clusters
and did the single cell resolution sequencing.
Long story, there was one gene
that was 256 times higher expressed,
and that was plaquoglobin.
And plaquoglobin here is part of the desmosome system,
cell-cell adhesion.
And if you look at the component of desmosomes
and adherence junctions,
where the cells are attached to each other,
this is the adherence, this is the desmosome.
Plaquoglobin is in both of those junctions.
Then we wanted to stain for plaquin,
which is only in desmosomes,
and cadherin, which is on the adherence junctions,
to see which ones are upregulated.
We developed the assays, and when we stain single cells,
these are single cells, clusters,
nuclear stain, tumor stain, cytokeratin,
plaquoglobin, desmoplaquin, and e-cadherin stain,
you see that only clusters express those.
Single cells did not.
Then we went to the patients,
or we did still more animal experiments.
We silenced the RNA for plaquoglobin,
and you see that the metastatic sites
were decreased significantly.
This is photon count.
This is the lung.
You see in the silenced case,
you don't have as many metastatic sites.
And it turns out to be when we do that,
number of cells were identical,
but the number of clusters decreased a lot.
Then we looked at actual patients,
breast cancer patients, and with the primary tumor,
we looked at the survival, metastasis-free survival,
and those patients that had high-expressing plaquoglobin
in their primary tumor and low-expressing,
there was a difference.
So you can see how important these things are
because you can see how you can develop therapeutic targets
by knowing these things.
And of course, we are working on a number of those.
Then one of the postdocs who's now a faculty member,
we said, we are seeing these clusters,
but it's like the early days of CTC chip.
We don't have the Ferrari version of the cluster chip.
We're just seeing because the chip
that's designed for single cell
sometimes captures clusters.
So we said, okay, can we build a Ferrari version
of a cluster chip?
And then we built what we call a CTC cluster chip.
It will be published next week.
So the idea here, unprocessed whole blood comes.
We have these individual triangles,
and the triangles are designed
with a very interesting mechanical feature.
It's not size.
This is about 12 micron here.
That's 12 micron, a two-cell cluster.
There's this apex here.
So if a single cell comes,
even if it is as big as this, it goes through.
And we have done the modeling.
For example, here, two cell comes, it's captured.
But if a two cell comes into a single hole of 12 micron,
it will go through.
It's viscoelastic.
So the way this works is almost like a cantilever.
The forces are balanced such that even a two-cell one,
and the flow rate is very small,
70 microliter per minute here.
So that's the concept.
It's not size.
It's the balance of the mechanical forces
such that even a two-cell cluster could be captured.
Then we build the chip.
It has eight parallel units,
eight rows, 4,000-some trapping triangles,
and I'm sorry, 70 microliter per hour, 33,000 triangles,
and 2.5 mil per hour volumetric flow rate.
And this is about the size of a business card.
And you see the electron EM images of the chip.
And look at it.
We could capture these clusters.
These are from patients.
Some of them are huge.
These clusters are circulating
in these people's peripheral blood.
This was a, we've seen this only once,
but it's so beautiful.
As much as it tried to squeeze by, and it did.
Look at this.
It went through that little hole, but it couldn't escape.
And we have eight rows.
So even if it escaped this one,
it will be captured in the next row.
And we have done all kinds of studies,
but what we now know is about 30 to 40%
of all patients, we see clusters.
Before it was 5% of the time,
because we were not going after actual clusters.
So it's a very interesting story unfolding
of the primary tumor goes to the circulation,
how EMT and cluster biology play a major role,
how some of these correlate with the patient outcome
and response to treatment,
and how these could be, of course,
potential target for treatments.
We are looking at the ways to, of course,
disrupt these clusters,
and we are looking ways to play with EMT
so that we can block the metastatic process
in using other approaches.
Clinically, the part that I didn't talk much today,
the way we see this technology play a role
at three levels.
So you have patients, they come, early detection,
patient group, treatment selection,
and real-time monitoring.
So the initial applications are in real-time monitoring.
Do they respond, do they not respond?
You look at the number of these cells,
you can monitor individually whether or not
the patient is responding to treatment.
And so this is more in the metastatic patient category.
The next level is we take these cells, sequence them,
look at the mutations, translocations,
so that they're, right now,
all drugs that are being developed for cancer
is targeted therapy.
They are targeting a genetic defect in the cell,
in the cancer.
There are no chemotherapies that are new
that are being developed.
All phase three, phase two, phase one trials are targeted.
So from what we call liquid biopsy,
you find these cells, sequence it,
you can tell which drug to give to a patient.
And so that's the near-term, next-level application.
Our most longer-term, exciting goal
is, of course, early detection.
Push the sensitivity of the technology
to a point where we could detect early.
That's the holy grail of cancer.
If you could do that, today's treatments
are cure for many cancer.
But the issue there, of course, is the sensitivity
and the presence of these cells early enough in circulation.
Then it could become a screening modality.
In that case, your false positives are intolerable.
You cannot look at millions of people
and have even tiny amount of false positive
and scare the hell out of them.
So the sensitivity of false positive,
you could miss it because other techniques miss it, too.
But you can't tell them they have cancer
if they don't have cancer.
And so those large-scale clinical trials
are extremely expensive, long time, take long time.
And you won't do it until the technology is ready for it.
In the very long term, we want to build,
from an engineering perspective, point-of-care chip
in the developed world for physicians' offices
where you give a blood.
Immediately, you could look at the presence of these.
But also a portable one for global health.
The data so far is actually such
that the number of these and the presence of these cells,
forget all the molecular fancy stuff,
correlates extremely well with imaging.
With MRI, CT scan, and PET scan,
even with the earlier, less sensitive technology.
So if that's the case, if we could reduce this
into a proxy for imaging in global health,
that could have huge potential.
So those are the areas that we are working on right now.
With that, of course, this is not done alone.
Dr. Haber is the Director of Cancer Center at Mass General,
is a very well-known cancer doctor and geneticist.
And we joined our laboratories,
so it is not a collaboration, it is a joint effort.
Whatever daily problem we have, we put our money,
post-docs, investment, everything in that problem to solve.
It's not in collaboration, it doesn't work that way.
More like a marriage, and Shamala is a brilliant
molecular biologist who oversees all of that.
Ravi Kapoor is a wonderful friend.
We started a number of companies together,
and he joined the team to make it work.
So in science, usually we do the discovery work.
We are very interested how things work.
But if you wanna have a real impact,
you need to make things work.
So Ravi is my most expensive date,
because making things work is very expensive.
Most of our money go to making those plastic chips
and making it uniformly at that cost factor and whatnot.
And that takes real amount, and Shannon does a lot
of the science behind engineering.
Leisha, lung oncologist, but sees all of our clinical trials
and IRB, pathology sequencing, and all the post-docs.
Most of the work I mentioned is by Fabio Murat and Emre,
the I-chip, and Fatih is the cluster chip.
But everybody played a major role.
Nicola here is the cluster person,
and David Miyamoto, all the prostate cancer.
And with that, of course, we get funding from NIH
to do the basic science, various foundations
to do specific cancers for those foundations.
We stand up the cancer to the multiscale clinical trial,
and Johnson & Johnson has the rights to the I-chip,
the third generation chip, and they will hopefully
commercialize it next year for researchers
on the antenna for in vitro diagnostics.
With that, thank you for your attention.
Thank you.

