Circulation April 21, 2020 Issue

Dr Carolyn Lam: Welcome to Circulation On The Run, your weekly
podcast summary and backstage pass to the journal and its
editors. I'm Dr Carolyn Lam, Associate Editor from the National
Heart Center and Duke National University of Singapore.


Dr Greg Hundley: And I'm Greg Hundley, Director of the Pauley
Heart Center at VCU Health in Richmond, Virginia. But I am
running to hide today, because I am going to get quizzed by the
master in the feature discussion. And listeners, it's really
interesting. It involves quantitative myocardial perfusion using
magnetic resonance imaging, but also adding the twist that
artificial intelligence computer algorithms are being used to
read the stress test images without any physician interference.
Oh, my goodness. I don't know what she's going to quiz me about.


Dr Carolyn Lam: Absolutely about all the AI algorithms and
exactly how you derive them. But why don't you tell us what you
want to describe first and the rest of the issue.


Dr Greg Hundley: Carolyn, I'm going to start with a paper on
peroxynitrite and as you know, that's a very short-lived free
radical produced in cells, part of the both oxidative and
nitrosative stress pathways. And this article comes from Dr
Swapnil Sonkusare from the University of Virginia School of
Medicine. Well, Carolyn, this study is involving mouse models,
and the investigators evaluated the relationship between
peroxynitrite, that powerful oxidative nitrosative stress
molecule, and obesity and hypertension.


Dr Carolyn Lam: Nice. And before you ask me anything more about
peroxynitrite, because I think you just summarize everything I
know. What did the authors find, Greg?


Dr Greg Hundley: What that found is that obesity induced
impairment of endothelial AKAP150-TRPV4 channel signaling
contributes to the loss of endothelial function and elevated
blood pressure. And lowering the levels of this oxidant molecule
of peroxynitrite reduces endothelial AKAP150-TRPV4 channel
signaling, vasodilation and blood pressure and obesity.


Dr Carolyn Lam: And what are the implications?


Dr Greg Hundley: Well, endothelial TRPV4 channels are essential
regulators of resting blood pressure and impairment of
endothelial TRPV4 channel activity contributes to obesity induced
hypertension. And therefore, therapeutic strategies, perhaps in
the future, that lower peroxynitrite levels can be used to rescue
endothelial TRPV4 channel activity, endothelial function and
blood pressure in obese individuals.


Dr Carolyn Lam: Nice, Greg. Well, I want to tell you a little bit
more about plain old hypertension. Now, we know that blood
pressure is regulated by the function of the kidney vasculature
and sympathetic nervous system, but do immune cells play a role?


Well, Dr Guzik from University Medical College, Krakow, Poland,
and University of Glasgow and his colleagues, studied the
relationship between major white blood cell types and blood
pressure in the UK Biobank population and employed a Mendelian
randomization analysis to examine which leukocyte populations
maybe causally linked to blood pressure.


Dr Greg Hundley: So we've got another blood pressure article.
What did they find?


Dr Carolyn Lam: They found potentially causal positive effects of
total blood lymphocyte count with blood pressure. Among the
mechanisms that might mediate this relationship, they found
evidence that blood lymphocyte count might influence albuminuria.
The study may additionally support a reverse, potentially causal
positive effect of blood pressure indices on blood neutrophil
monocyte and you sit a full count. So fairly interesting.


Dr Greg Hundley: Very nice. So I'm going to switch over and talk
a little bit about lifestyle interventions. I know you're a big
fitness buff. So this paper is about fitness, body mass index and
the risk of heart failure in overweight, obese individuals with
type two diabetes mellitus. It's an analysis from The Look AHEAD
Trial. The corresponding author is Dr Ambarish Pandey from the
University of Texas Southwestern Medical Center. And a little bit
of background Carolyn. Type two diabetes is associated with a
higher risk of heart failure and the impact of a lifestyle
intervention and changes in cardiorespiratory fitness and body
mass index on the risk of heart failure in this population is not
well established.


So what are the authors do? They had 5,109 participants from The
Look AHEAD or The Action for Health in Diabetes Trial, without
prevalent heart failure at the time of their inclusion. They
implemented time to event analysis to compare the risk of
incident heart failure between an intensive lifestyle
intervention versus a diabetes support and education group. The
association of baseline measures of cardiorespiratory fitness
estimated from a maximal treadmill test. The participants, BMI
and longitudinal changes in these parameters with the risk of
heart failure were evaluated using multi variable models.


Dr Carolyn Lam: Wow. What did they find?


Dr Greg Hundley: Carolyn, very surprising. Among participants
with type two diabetes mellitus. In The Look AHEAD Trial, the
intensive lifestyle intervention did not, did not appear to
modify the risk of heart failure. What they also found is that
higher baseline cardiorespiratory fitness and sustained
improvements in cardiorespiratory fitness and weight loss were
associated with a lower risk of heart failure.


So even that education group where patients started doing things
more, had more cardiorespiratory fitness and baseline and sustain
that with weight loss, those were the ones that had lower risk of
heart failure.


Dr Carolyn Lam: Nice summary. Well let's go through what else is
in today's issue. There's a research letter by Dr Eitel on the
impact of morphine treatment within without metoclopramide
co-administration on ticagrelor-inducted platelet inhibition in
AMI and that's the randomized MonAMI trial. There's also an
in-depth paper by Dr Eijsvogels on exercise and coronary
atherosclerosis. So, this interesting review describes the
effects of physical activity and exercise training on coronary
atherosclerosis in middle aged and older athletes and really aims
to contribute to the understanding of the potential adverse
effects of the highest doses of exercise training on the coronary
arteries. Very interesting.


Dr Greg Hundley: And Carolyn, I've got a perspective piece and it
comes from Dr Robert Stravitz, as well as Dr George Vetrovec from
VCU, and it evaluates the risk of invasive cardiac procedures in
patients that have liver cirrhosis.


Then finally there's a very nice ECG challenge. It's the anterior
STEMI without S T elevation in lead one and it comes from Dr
Yun-Tao Zhao from the Peking University Aerospace School of
Clinical Medicine. Oh no. Oh, I've got to run.


Dr Carolyn Lam: Now, we're going straight into the prolonged Dr
Hundley quiz.


Dr Greg Hundley: Very good.


Dr Carolyn Lam: Oh boy. Today's feature paper is really a biggie.
It talks about AI and its approach in quantitative myocardial
perfusion by cardiac magnetic resonance imaging. Now, I know
we've been building up to this discussion already right from the
start because I've got my beloved cohost, Greg Hundley talking
about this, but wearing the hat as the editor who managed this
paper and also so pleased to have with us the corresponding
author, Dr James Moon from University College, London and Barts
Heart Centre in UK, as well as from across the other side of the
world. We have Dr Peter Kellman from the National Heart, Lung and
Blood Institute of NIH. So welcome gentlemen, what a great paper.
I think in Greg's words he said earlier just really landmark. I'm
almost paradigm shifting in this area. So let's dive straight
into it. And so maybe James, could you start by telling us how
does CMR quantitative myocardial perfusion usually work in
today's world without the AI and perhaps you know what the study
showed that it can do. What can AI help us do?


Dr James Moon: If we want to understand where chest pain comes
from, we can sometimes use cardiac CT, or we can use a functional
test and a different function test. But a particular test with
bandages is chronic MRI. And what we do is we give a dye into a
vein and track that through the heart. And the upslope can tell
us about a fusion. And if you do that during vasodilators stress,
you can see regional differences visually. Now, the interesting
thing is that if you take those signals and using a team such as
Peter Kamina, NIH, you can make that constative and, in fact, do
that inline on the scanner, so you get color maps where the
pixels of valid in mils per gram per minute. So you can see
exactly how much facilitation there is. And what we share in here
is that if you quantify that, literally, automatically on the
scanner using artificial intelligence approaches from the NIH,
those values are incrementally prognostic predicting outcomes for
patients.


Dr Carolyn Lam: Wow. Okay. So give us a little bit more here,
James. What do you mean by incrementally prognostic? Did you
compare it with the best human readers and perhaps have a subset
and I think you did, of course, who the human readers can see
something, and AI did, right? Could you tell us a little bit more
about that?


Dr James Moon: We've been doing a lot of the sort of technical
development papers after the common initiatives from the NIH. So
we've been doing the correlations across a number of sensors
with, for example, pets and animal models. And really what we did
here was we overlaid our clinical service with the AI and in more
than a thousand patients actually at two sites, we were able to
follow up patients and a number of those of course, and sadly
underwent an event, death, or mace. And what we are able to see
is that the stress flow and the ratio stress to rest flow, were
independently associated by death and base and mace. So for each
mil decrease in stress flow that has a great share went up. And
that was as you say, something that you just can't do viscerally
because you can't get that global background change.


Dr Carolyn Lam: Oh this is scary. So AI beating out humans, I
suppose is what you're saying, not quite maybe. Or is it right?


Dr Greg Hundley: You know, we have to do a number of things in
health care and we're often drawing circles, drawing consoles,
identifying regions of interest and these are quite mundane tasks
and if you can train your AI correctly, then that can do that for
you, leaving you and freeing you up to higher executive
functions, like discussing the results with patients. And I
wouldn't be scared of the AI here because the way PISA
implemented this, the contours are all drawn, so it's completely
explicable.


Dr Carolyn Lam: I really liked that explanation. And so what most
people fear when AI comes about is, they fear this black box
approach that they cannot understand. And what you're saying here
is it something understandable and it will facilitate efficiency,
which is beautifully done, and what you've shown as well. So
Peter, it sounds to me that it took a long time to develop this.
Could you describe what it takes to get an algorithm and AI
algorithms like this?


Dr Peter Kellman: This is quantitative perfusion and perfusion in
general using cardiac mechanisms has been in development by a lot
of labs around the world over the past 20 years. And the specific
objectives we had were to get it out of the laboratory into the
hands of the clinicians and do that, we chose to implement
everything fully, automatically and integrated onto a clinical
scanner so that the clinicians would get the answers on the fly.


That is to say they would perform a scan and it would be
transparent. They'd run the scan the way they normally do today,
and the results of the myocardial blood flow would appear on the
scanner within a minute. So, we probably worked on this, myself
and Dr Wie Wzei at NIH, for the past five years. Gradually
building this up to the state where it is now at sites like The
Barts Heart Centre can use it.


Dr Carolyn Lam: Just really hats off. This is amazing because
you've summarized it in a few sentences, but you know I kind of
personally know what it takes. Just getting an algorithm is one
thing but making it clinically usable and then James
demonstrating it clinically are all huge achievements. So
congratulations. I don't think anyone can say it better than
Greg. Greg, tell us why is this paper so important? And see that
was an easy question.


Dr Greg Hundley: This is really dramatic in the field of
cardiovascular magnetic resonance, but also for stress testing
related to patients with cardiovascular disease. As many of our
listeners know today, the majority of studies are interpreted
visually and while the spatial and temporal resolution is just
beautiful like high definition television of the inside of our
bodies, the artifacts that sometimes occur and then some of the
interobserver variability remains problematic. What I love about
this, is this is now a quantitative. It's not a visual
interpretation of signal intensity. What both Dr's. Kellman and
Moon have developed is something that measures the blood flow in
mls per minute, per gram of tissue regionally throughout the
heart and then color codes that so that all of us can interpret
it readily. I think another big piece of this is that they have
created the artificial intelligence that helps interpret that for
us and just think now we'll be able standardize readings across
medical centers.


That's an enormous advance for this field and combining this
information and putting it into, both national as well as
international registries, could be very important for identifying
abnormalities that would forecast prognosis. And these gentlemen
have laid the framework for that because they have a large number
of subjects, two centers, and they had comprehensive follow up
that looked at heart events and this technology was able to
forecast that on the backdrop of all the other parameters,
demographic variables, risk factors, etc., that we find in
patients that present to us with cardiovascular disease and are
symptomatic with chest pain syndromes. So boy, what an
outstanding imaging paper and feel so fortunate to actually have
this in circulation.


Dr Carolyn Lam: Greg, no one can say it better than you. I would
like to emphasize that this just doesn't hold the implications
for MR imaging. I think it holds implications for many forms of
cardiac imaging if we have AI to assist us. So James, how do you
see this helping the patients? These are prospectively performed
cardiac MR scans, but you know, what about surveillance of MR
databases? What are the implications for perhaps detecting
microvascular abnormalities? Another pet topic on my own. If you
don't mind, you know, tell us, what do you think are the clinical
implications?


Dr James Moon: Quite a lot of clinical implications. We're really
starting a journey here. So, one of the things I'm interested in
is that we have been in an era of thinking about the epicardial
cone use only. We were seeing significance between patient
differences in peak hyperemia related to, for example, age and
presence of diabetes, but must be reflecting the microvasculature
and of course, that may be a key biomarker for the future and
understanding what happens to patients. So naturally, of course,
when we're assessing the epicardial arteries, we're going to have
to understand that those assessments, especially if it's
measuring wall flow, will be influenced by the microvasculature
because it's a circulation. So, this may just bring that
microvasculature into the mainstream as potentially therapeutic
target.


If you think about this AI and observe a variation. So what we do
is measure things clinically and that sounds like boring old
science, but metrology and accuracy and precision translate into
something that cascades down. So that key dots the patient
relationship where we're making the decisions on what therapies
together might influence outcome. And if you can rely on those
results and if the results that scanner one in one place and
another scan at another place always read the same, then your
healthcare system just gets a boost because everything is more
reliable.


Dr Carolyn Lam: That's beautifully put. And Peter, thank you to
you and your group for developing such AI and James for
demonstrating this. Now if I may, this is an amazing
collaboration. I understand the British Heart Foundation had a
role in supporting this. How did you get together to decide to
work across regions and so on, on this? Could you just give us a
little bit about that?


Dr Peter Kellman: Our working together was an outgrowth of prior
collaborations in areas such as T one map and for tissue
characterization and Barts Heart Centre represents the largest
cardiac MR site in the world, in terms of volume. And when you
have a quantitative method, it was the perfect place to evaluate
this kind of ischemic heart disease. And furthermore, when you
talk about AI, you need to train the algorithms and develop the
models. You need a large volume of data. So, right now we're
probably doing on the order of 150 stress tests a week. And so,
after a year we had a large volume of data to train the
algorithms for segmenting the myocardium. So the collaboration
has been ongoing for about five years.


Dr Carolyn Lam: That's great. Greg, you know what, because I
adore you, I'm going to ask you to give us the final words. Tell
us your thoughts on where this is all heading.


Dr Greg Hundley: I think Peter astutely identified that Barts
Heart Centre and you're seeing really in the United Kingdom the
primary use of magnetic resonance imaging for cardiovascular
stress testing with its accuracy. And now, adding to this these
quantitative assessments that can reflect, not only epicardial
disease, but as we've heard, have the potential to identify micro
circulatory disease. I see this growing and extending worldwide.
And I think the next studies will involve the use of this type of
technology and multiple different manufacturers of scanners and
across different field strengths, 1.5 and three T, and
multi-center initiatives using this technology to try to forecast
cardiovascular prognosis in patients that present with chest pain
syndromes as well as other disease processes that involve both
epicardial coronary arteries and micro circulation.


Dr Carolyn Lam: Wow. And we're so pleased and proud to be
publishing then this paper here in circulation this week. Thank
you so much, gentleman. It's been awesome discussing this with
you.


 And thank you, audience for joining us this week. You've
been listening to Circulation on the Run. Don't forget to tune in
again next week.


Dr Greg Hundley: This program is copyright, The American Heart
Association 2020.