Subscribe
Share
Share
Embed
Proteomics in Proximity discusses the intersection of proteomics with genomics for drug target discovery, the application of proteomics to reveal disease biomarkers, and current trends in using proteomics to unlock biological mechanisms. Co-hosted by Olink's Dale Yuzuki, Cindy Lawley and Sarantis Chlamydas.
[Music]
Welcome to the Proteomics and Proximity Podcast.
We are co-hosts, Dale Yuzuki, Cindy Lawley, and Sarantis Chlamydas from Olink Proteomics
talk about the intersection of proteomics with genomics for drug target discovery,
the application of proteomics to reveal disease biomarkers,
and current trends in using proteomics to unlock biological mechanisms.
Here we have your hosts, Dale, Cindy, and Sarantis.
Hello, everybody! Welcome to our episode dedicated to biomarkers of heart failure.
I welcome with me today Dale and Cindy. Hey. Hey there.
And today the paper that we will discuss is about protein biomarkers of new onset heart failure,
and is published in the Circulation Heart Failure journal.
The first author is Nicolas Girerd, and the last outer is Professor Faiez Zannad.
The most important finding of this paper is their use proteomics and Olink proteomics
in order to identify biomarkers related to heart failure,
and it is a great, it is an amazing study
looking in three independent cohorts, ARIC, FHS and HOMAGE,
and finally it was dedicated to the
secretome, proteome, secretome, and biomarker discovery connected to heart failure.
Cindy, would you like to give us a little bit of background of these cohorts and some of the
characteristics of these cohorts that they have studied? Thank you.
Sure, absolutely. I'll just summarize them in the order that they are in the title.
So the heart, omics, and aging cohort is the HOMAGE consortium,
an EU-funded program, and they're looking for personalized strategies for
understanding cardiovascular disease. This is a consolidated effort among eight European countries.
I mean, this is just a massive undertaking, right? So they have eight different countries looking at
heart omics, and aging, so the two are together? So both, yes, together. But within this subset,
there's three represented groups. So there's something called predictor, health ABC, and prosper.
And that's from subset of samples from within 20,000 that are this combined
consortium that represents HOMAGE. Yeah, when you start digging into
these cohorts, it's just awe-inspiring what is able to be done bringing collaboration together,
right? So, HOMAGE is roughly 20,000 large or even larger? Oh, it is. That's right, 20,000 large.
Yep. And then there's a subset that is represented within this study. I see.
So I think the HOMAGE, I think they had 562 individuals in this study that were cases.
And again, those were identified, as Sarantis mentioned, incident heart failure. So,
the characteristic was that they were diagnosed and hospitalized, but not necessarily,
or they would have been excluded if they had been identified with heart failure coming into
the study. So the importance there is to be able to more likely identify the markers that are
linked to predictive value in predicting heart failure, the likelihood. And we're trying to,
you know, tease out the pathogenetic biomarkers, right? Over time, with these studies, we're trying
to understand what's predictive and what's incident, what happens after you have this diagnosis.
So for the HOMAGE study, then they ended up taking healthy and non-healthy individuals of a certain age
and then follow them over time? Yeah, that's right. All three of these are longitudinal studies.
As, you know, my understanding is that the exclusion criteria was if they had heart failure
in their diagnosis in advance that they were excluded. So here, they're really trying to get at
incident heart failure. You know, what are the predictive markers that help us identify
that heart failure is going to happen? And so as far as it skews older, right? So we're getting,
that's right. We're getting an older group of individuals, regardless, I mean, meaning they're
nominally healthy, but oftentimes they have just any different kinds of conditions other than heart
conditions? So it depends upon the cohort, right? So the other two cohorts, as Sarantis mentioned,
are the extremely famous Framingham Heart Study, right? And again, that was a community study.
It was based on residents in Framingham and that group is actually in their seventh exam. That has been
going since 1948. So I remember when they were first using genetics in that study and some of those
first publications that came out there, there's some pretty impactful stuff that comes out with Framingham.
If I understood correctly, chances are that there was not any diagnostic method, right? At the time,
the point zero is entering the hospital with heart failure, right? If I understood correctly,
and then they got diagnosed with heart failure. For the subset of samples in this study, right?
But not the full Framingham study. Okay. But yeah, so in Framingham, there were 191 cases and a matched
set of controls. So they would refer to this as a nested case control design because it's nested
within the larger cohorts that these samples were identified and then matched. And again,
what a phenomenal resource to be able to have so many cases and so many controls that are so many
individuals in a cohort to be able to match those controls as well as they were able to. And then
the third cohort that you mentioned was the ARIC study. And this, I think, is really exciting
included in here because it's over 30% of African diaspora individuals. So individuals that
are declaring their ethnicity in the study as black, which I think this is incredibly important for
us to understand what we've been missing by the over-representation of northern European populations
in studies around heart disease. And the data for heart disease among the black community is
much higher, right? As far as heart failure goes? Yes, that's my understanding and certainly,
you know, we have lots of findings that we would never have seen. PCSK9 is one example of them
that we wouldn't have seen had we continued to look at European populations. So this,
you know, this just double clicks on that. What we've said several times is this importance to not lag
behind in understanding omics around these important diverse populations as we go forward.
You mentioned something about the Framingham cohort that this was the seventh examination.
Right. So it's actually the kids, the offspring of the originals.
That's what I was getting at. 1948 is getting pretty old.
Not many people are still around.
Yeah, that's a very important... Yeah, the progeny of that original group...
Framingham heart study then is on the second generation of the original people.
Is that correct? Or did they take... That's my understanding.
So they have the genetics of the parents and now they're looking at the offspring.
Yeah. Wow. And then the seventh examination was what 1998 to 2001 so that they were
taking blood samples and full medical workup of all the children.
Yeah, that's right. And actually there's a comment in the paper about the Framingham
participants being a little younger than the other two cohorts.
By about 10 years, I think.
I'm trying to imagine what it would be like to be the son or daughter of a Framingham volunteer.
It's like, mom, you did what? I have to do this because you signed up in 1948?
Well, are they doing the next generation after that?
Yeah, it's a good question. I don't know.
I think there is. It's one of these... There's so much interesting data, right?
Yeah. And having them nested within families tells us so much about what's passed on
from parent to offspring, and genetics we think of as kind of static and then these other biomarkers
dynamic over time. I think if we could just run all samples in that Framingham Heart Study from
the beginning to now, imagine what we might learn just with the broadest look at the proteome.
Speaking of which, let's talk about the proteins that were looked at in this study.
That's a good point. I mean, I would have expected to look a little bit broader to be honest in
something that they mentioned in the Discussion [section], right? That they should have gone more in the
exploratory. I think they used the allocated panel. I think that was what was available at the time.
Right? So over time, you know, just like I always compare these genetics, you know, like my frame
of mind is genetics, right? But over time, you know, the first broad scale tools to look at SNPs were
around 1536, right? So being able to expand that to a million or five million on these chips
that are available today is just a representation of advances in innovation. And I think we
seen that obviously, us here at Olink in proteomics. So I would love to see these, a publication around
an even broader look at the proteins. And I think with what they've found here, especially with the
value of this cohort, that perhaps that's something that they're thinking about, it would be great to
talk to them and ask them. Yeah, I think the [Olink Explore] HT [proteomics platform] would be great to be able to do that.
That's right. Yeah, you want to tell our audience a little bit about our latest, broadest look at the proteome?
I think Dale would be the best person to answer that. Yeah, he's actually developing all the content.
I'd really interested, Sarantis and Cindy, to hear your sort of input on this, right? Because I was a
person behind the curtain rolling it out. I'd like to hear both of you describe it in your own words.
Okay, Sarantis, you first. I mean, I would say it in a very simple word. I mean, it's like a
revolution, right? Of high-through proteomics, right? This is the ultimate tool and
assay to go exploratory. We offer like more than 5,000 biomarkers that were selected from artificial
intelligence, from the public research, from the opinion leaders, voice of customers. And now we are
really at the age of proteomics. And then we offer this great tool for people to explore and
identify biomarkers in many digits. Now, I think it's the ultimate assay in the field of proteomics.
And also we offer it in a very high-throughput and efficient manner.
It's really adjustable to the new generation sequencing capabilities, right? And this is like
makes it the ultimate platform. I would say that's the I'm really proud. I'm really proud. I'm looking
forward to see the first data that comes out of this platform. Not that I see it
from the outside. But that's great. Cindy, you want to add to that?
Yeah, sure. So what I'll add is that one of the elements that Olink has
integrated into this new product is the ability to ease the workflow and the way we've done that
is by, instead of having a very diseased focused organization of the panels,
And when I say panels, I mean, you know, subsets of proteins that make up, for example, our
Explore 3072 [platform], rather than being disease focused, we instead are focusing - And we've talked in here
before about how many proteins we cover in the low abundant proteome. So those are ones where we're
measuring them in a neat fashion, meaning one-to-one, we don't dilute them at all in order to
discriminate between diseased and healthy individuals. And so we've organized them, rather than organizing
them by cardiovascular, oncology, inflammation, or neurology, we're organizing them by dilution
block, right? And so we've been able to really streamline the workflow as a result of that. And
groups that I'm talking to that are doing large numbers of samples in this population health space
can triple their throughput or their capacity with a very modest investment in automation.
And of course, you know, at this scale, we're an automated platform, controlling beginning
to end the processing of the samples in an Olink Explore environment. That said, I would not expect us to
sit back on our laurels in trying to cover as much of the proteome as we can in this targeted fashion.
Of course, we launched the the [next-generation sequencing] NGS platform in 2020. We doubled the number of proteins in 2021.
And now we're almost doubling it again. So it's like you say, Sarantis, I'm very excited and
very proud of of what our R&D team has been able to accomplish in a very short time from my
perspective. And building for that scale, even this particular paper, right? The HOMAGE cohort had
560 cases and 870 controls. That's 1300 samples right there. And people's heads are
just kind of like, wow, that's a lot of work to do, right? 13 micro-titer plates. And then ARIC had
250 cases, 250 controls. That's another 500 on top of the 1300. And the Framingham had about
400. And so that's, okay, what are we now at? We're at 1400 plus 500 is 1900 plus 400 is, you know,
2500 samples, right? Just say that's a very - that's a lot of work for someone. And yet for the
Olink Explore HT, right? It's built for this kind of scale in terms of scaling the number of
samples and being able to finish projects really efficiently with an NGS readout. So super exciting,
right? In terms of what our customers and what the scientific community will be able to discover.
With this particular paper, it was only using Three 96-Plex Olink Target 96 panels. And you say,
well, it was only ten years ago, it was amazing to do 260 to 270 proteins
at a time. But then you think, well, we'll up that by a factor of 20. They go from 270 to over 5,300.
Then it's just, wow, what can you do with 20-fold times the biomarkers?
Yeah, or the potential biomarkers to then narrow it down to the ones that are the pathways
of interest, right? Because that's the dream. The dream is to be able to make discoveries as broadly
as possible and then narrow down to a panel, a subset of proteins that are going to really improve
your ability to predict heart failure above and beyond clinical factors being used today,
like NT-proBNP that they talked about in this paper. Yep. And that is what the whole
goal of the research was, right? The whole goal was to improve prediction. So, Sarantis, how can you
comment then on what they found? I mean, they found 142 proteins, like potential biomarkers connected
to heart failure disease. Of course, in order to be classified as biomarkers, they need a
lot of proteomic studies to follow up. They mentioned that they don't know
if some of the proteins that are caused by or affect heart failure, right? They need to do some
studies in mouse or probably some follow-up studies. But the important thing that this,
among these 142 proteins, there are eight proteins that are common in all of these
three cohorts. And among these eight proteins, of course,
natriuretic peptides were there, something that was expected to be. And some that
pop up to my mind and we were discussing before, it was the eukaryotic translation initiation factor
4E-BP1 protein. It's a really interesting protein because it's involved in stress and
metabolic stress and actually binds at the 5' of mRNA and blocks the translation
of mRNA when we have a metabolic stress. And it's also regulated by mTOR. And I know that
David had a great story about tropomyosin. And I think it would be great because mTOR
is a target of tropomyosin, and tropomyosin is a really famous drug.
In the aging space, yeah. It was about a year ago, I was at a user group meeting
for Olink. And a professor from Institute of Systems Biology named Nathan Price was there
and gave a really interesting talk about the early or the wellness and the whole markers of wellness
and the markers of sort of illness and that transition phase. They're mining data from a startup
company called Arivale from years ago that they published several different papers using Olink
technology and they continue to. Well anyway, about a few months later, he and Lee Hood,
the founder of ISB, published a book called, "The Age of Scientific Wellness." And so, me being curious,
I wonder if there's Olink data in this particular book on longevity and wellness.
And I went ahead and read it, and it just blew my mind, right, with the different research on
sirtuins on mTOR, and mTOR in particular was interesting because they talked a little bit about
rapamycin and the ability of rapamycin as sort of an anti-aging longevity compound. I mean,
I'm like, what? I remember reading about mTOR as being this sort of master regulator, or what have
you, but to include it in aging? Well, fast forward here to just a few months ago,
and I'm listening to a podcast by Paul M. Cooper called, "The Fall of Civilizations." And it's about
history and it's about different places and different peoples that I'm not very familiar with.
And there was one podcast on Rapa Nui, which was Easter Island. And they basically call it a
collapse or a contact between civilizations. But it was on Rapa Nui that rapamycin was isolated from.
It's inside a volcano on Easter Island in the Pacific Ocean, where scientists collected an
unusual specimen, right? They thought, okay, well, this looks like an antibiotic in terms of the
way it was growing. So they went, took it back. I think it was a Canadian group that ended up isolating
rapamycin. And then naturally, right, the biology of it, mTOR is the mammalian target of rapamycin.
And it's completely off-label, right, for people to start taking rapamycin just for longevity
benefits. It's a crazy world once you get into the whole longevity, aging, you know, wellness
of business. But the science is rapidly catching up to do away with all charlatanry
that is going on or has gone on. But nonetheless, when I saw, you know, 4E-BP1 in this
particular paper as being a significant biomarker and then, Sarantis, you tying it back to mTOR,
what mTOR being an effector, right? A regulator. Yeah, it phosphorylates.
Yeah, phosphorylates 4E-BP1. Yep, biology is a system. And it's all connected.
And as far as these biomarkers, though, what can you tell me then about, Sarantis, about the C-index
that they refer to in the paper? Yeah, C-index is
similar to AUC, the area under the curve. And it actually defines the
prediction, actually, how accurate it would be a prediction for the model, right? And the
really important thing that is here, and it's a fact that when they compare or when they add this
proteomic signature to the clinical model, they see a delta C-index increase, and that was because
there was more accurate the model. And actually it was increased even more,
compared to when they add the typical factors, the typical biomarkers, like the
naturietic peptides to the clinical factors, right? And there was an even larger increase.
That means that adding proteomics to already existing, let's say, diagnostic tools enable
and make the diagnosis a little bit more accurate than before. At the level of, I see the
C-index increase from 5.9% to 11.1%, that's clearly significant.
So it's a performance metric. Yeah. It sounds like, yeah, that's really helpful. Yeah, because I was
looking for AUCs on here, and I see the delta C-I. The C-index, right?
Yeah, looking at figure three, they show the increase in the C-index from what I understand.
And concordance index, right? Yeah, okay. And it's roughly an increase of 10% from 77 to 88,
from 73 to 79, from 74 to 81. So these are pretty nice jumps in the C-index value when you add the
biomarkers. Now, a simple question, was it a lot of biomarkers? Or was it just the
nine that they identified? No, there was these biomarkers, they were selected among these
142 that they were identifying three cohorts, right? Oh, so it's all 142? I mean, they say that the
excluded some of them in the analysis, but it was among these 142 for each of these cohorts, because you
see that they have classified in figure three, they have classified the three cohorts. Yeah.
And they have seen what is in each cohort, what is the difference, right? And then there are
specific biomarkers in these cohorts. Yeah, I see. And they pointed to the inflammatory
pathways as well as the remodeling pathways. Now, they call them mechanistic clusters related to
these broad categories, which, you know, inflammation makes a lot of sense. I think this idea of
extracellular matrix and apoptosis, I think that sort of points back to your
linking this to mTOR, Sarantis. And remember, we talked about the EMPEROR study in a
previous podcast, and some of those pathways they commented in their paper might be linked to some of
these, you know, sirtuin pathways. So, I will say the complexity of proteomics is broad.
And I find these anchors that I'm trying to hold onto to start to understand
the mechanistic biology behind some of these diseases that are, frankly, the body is so good at
using these chemicals in our various different organs, we don't have, you know, one smoking
gun in any given disease. And so starting to try and put these things together is going to require
some machine learning and some ... So this idea of predicting heart failure,
you have underlying, I mean, the first take-home was inflammation, right? The inflammation,
TNF and other inflammation-related proteins and interleukins. The other big take-home was
remodeling, right? And this is what Cindy is talking about in terms of even before a person
has heart failure, there is apoptosis happening. There is hints around autophagic flux, which is
the SLG-T2, the Emperor study points out. And then, of course, is the possibility of
this is with the lens of [analyzing] only 270 proteins, right? They only use three Target 96 panels. I say "only,"
but still a lot of proteins. And still what, 2500 samples, so still a lot of samples,
and some significant findings, which is what makes this paper so tantalizing, right? To say,
okay, we've got a very good snapshot of inflammation. We have hints on remodeling, right?
As well as, I think one of the things that they ended up - just remembering the figure we're just
talking about - they're looking at all the protein specific to the cohort. It made me think,
yeah, well, perhaps there just was enough, like, high signal when they narrowed it down, right? They
lost that impact of predictive power. But nonetheless, this is the first snapshot. Any other good take-home
messages here? I would say, you know, they talk about the promise of predictive
tools in order to pave the way for design of predictive or preventative trials, right? Like
the vision of being able to have these preventative trials, especially when we have imperfect
therapies for heart failure today, which I really appreciate them clicking on that. That's
one of the last points they make before they go through some of the limitations that
Sarantis commented on earlier. So that, I think that's exciting. The world of prevention, right?
If only. Yeah, so many different ways. So I've gone on this sort of hobby now of
reading a lot of wellness and longevity books, really, really interesting stuff. Why? Because the
science is catching up, right? And the science is now looking into longevity and aging as
a separate discipline. So maybe we'll talk about the future. I was just noticing, yeah, I was just
noticing that that Peter Attia's new book is now on the top 10 bestsellers list. And I was mentioning
that in a group at a conference where I was at in Melbourne, a group of four people who were standing
around saying goodbye, you know, good meeting, heading to the airport. And I mentioned that book
and somebody reached into their backpack and pulled it out. It's funny you mentioned that book
because I just finished it. It's called, "Outlive." Awesome. Highly recommended.
Peter's podcast is fantastic as well. And he talks quite a bit about rapamycin as well.
Yes. Now on that note: Hey, great to see you all. Thank you. That was great. Take care.
Hope you're having a good summer. Enjoy your summer. Take care. All right. Bye-bye.
Thank you for listening to The Proteomics in Proximity Podcast brought to you by Olink
Proteomics. To contact the hosts or for further information, simply email info@olink.com.