CVM Grand Rounds November 19, 2025

Name: CVM Grand Rounds November 19, 2025
Uploaded: 2025-11-20T13:35:34.5166667Z
Duration: 1 h 16 min 5 s

November 20, 2025

Information

ID: 13645
To Cite: DCA Citation Guide

Download Transcript

00:01No.
00:05I'm just making sure everything's
00:06working.
00:09Eric, can you grab your
00:10ear?
00:11Well, I'll sit next to
00:12you. Oh, I'm just gonna
00:13call back to the wall
00:14pen.
00:28Yeah.
00:42Yeah.
00:55Alright.
00:56I think we need to
00:57get started with Grand Rounds.
00:58We've been running a little
01:00bit late on the Symposium.
01:03So people are out, wandering,
01:05looking at the posters, and
01:07they're coming in now.
01:10But I can
01:12so to start with some
01:13general announcements,
01:14so,
01:16this is
01:18shown on the screen is
01:19the code for,
01:20getting, CME credit for cardiology
01:23grand rounds.
01:26The afternoon session will have
01:28also,
01:29a Zoom link to get
01:30additional CME. So the whole
01:32symposium
01:34is offering,
01:35four point five,
01:37credits CME credits.
01:42Shown here is the upcoming
01:44cardiovascular
01:45grand rounds.
01:48In the beginning of December,
01:49we have for peripheral vascular
01:51case conference.
01:53And then, middle of December,
01:55there's a a faculty research
01:56meeting. And then on the
01:57seventeenth,
01:59there's the meeting required annual
02:01bill billing training.
02:06And, of course,
02:07our,
02:09Grand Rounds speaker is,
02:12Todd Valines,
02:13and,
02:14this is part of sort
02:16of giants and Yale Cardiovascular
02:19Medicine. And this grand rounds,
02:21in addition to this symposium,
02:23is in honor of
02:25Barry Zarett.
02:26Excuse me.
02:30These are the disclosures.
02:34So I wanted to start
02:35off and just say a
02:36few words about Barry.
02:38So,
02:40as many of you know,
02:41Barry
02:42was born in New York
02:43City, grew up in Brooklyn
02:44in the Queens. He went
02:46to Queens College in New
02:47York City,
02:48went to NYU School of
02:50Medicine, did his residency and
02:52internship at Bellevue Hospital,
02:55then a cardiology fellowship at
02:57Johns Hopkins University and was
02:59in the Air Force and
03:00ended up being,
03:02positioned at Travis Air Force
03:04Base where he did sort
03:05of pioneering work,
03:08in in the cardiovascular
03:10space.
03:12It was in nineteen seventy
03:14three that he joined the
03:16faculty at Yale,
03:17and shortly thereafter he was
03:19appointed as the chief of
03:20cardiology,
03:21and then a professor of
03:23medicine and radiology in nineteen
03:24eighty two and became the
03:26Robert w Berliner professor of
03:28medicine in nineteen eighty four.
03:31Barry has many accolades.
03:33He was an established investigator
03:35with the American Heart Association.
03:37He was a member of
03:38the American Society of Clinical
03:40Investigation,
03:41and importantly, he was the
03:43founding editor for the Journal
03:44of Nuclear Cardiology,
03:47which he served for ten
03:49years.
03:50He served as a medical
03:51director at the hospital
03:53and, and was the chief
03:55of cardiology here,
03:57from nineteen seventy eight to
03:58two thousand and four
04:00and was also associate chair
04:03of clinical affairs in the
04:04Department of Medicine.
04:08But as many of you
04:09know,
04:10Barry
04:11was recognized
04:12as a truly gifted clinician,
04:15a teacher, a researcher,
04:17and a mentor to many
04:18of us.
04:20He was a true pioneer
04:21in nuclear cardiology.
04:24Initial work he did was
04:26was in measurement of coronary
04:28blood flow,
04:30with the intracorone injection of
04:31radioactive
04:32xenon.
04:34He pioneered,
04:35the work initial work in
04:37myocardial perfusion imaging
04:39with, development of potassium forty
04:41three imaging, which then was
04:43replaced by thallium to a
04:45one imaging.
04:47He also was instrumental in
04:48moving forward gated blood pool
04:50imaging.
04:51And he
04:53really was the one introduced
04:54the use of assessment of
04:56ejection fraction
04:57to actually guide and monitor
05:00patients who are on chemotherapy,
05:03to prevent,
05:04cardio,
05:05cardio chemotherapy induced cardiotoxicity.
05:09He had an important role,
05:11as part of the Timmy
05:13executive committee
05:15and was a principal investigator
05:17and ran an important core
05:18lab for many years here
05:19at Yale and a lot
05:21of sort of the infrastructure
05:22that was developed as part
05:23of that remains in place,
05:26for other,
05:27course.
05:30He sort of advanced
05:31first pass angiography, which we
05:33don't do very frequently right
05:34now,
05:35and that included actually ambulatory
05:37assessment of cardiac function including
05:39during mental stress.
05:42He was involved in the
05:43evaluation of many new tech
05:45labeled perfusion agents
05:48that involved first in human
05:49studies.
05:51And importantly, he sort of
05:53made a major shift in
05:54the whole area of nuclear
05:56cardiology and promoting,
05:59molecular imaging amongst the community.
06:02He's published,
06:03he's well published and he's
06:05best known for,
06:07the the textbook that he
06:08co edited
06:10with George Beller on clinical
06:12nuclear cardiology.
06:16So shown here are,
06:18two of the classic images
06:20from Barry's, sort of pioneering
06:22work in blood pool imaging
06:24and potassium forty three imaging.
06:32So as you know, Barry
06:33was an avid runner, a
06:35painter,
06:36and a family
06:37man and, and he saw
06:39cardiology
06:41and the faculty as his
06:42extended family.
06:44We had, annual picnics at
06:46his house and softball games
06:48and, and he was sort
06:50of, always the pitcher at
06:52the softball game and trained
06:54for weeks in advance in
06:56preparation for the softball game.
06:58His sons,
07:00you know, he recruited to
07:01help the faculty,
07:02so to beat the fellows.
07:07And so in his honor,
07:09we have the special cardiology
07:11grand rounds and we have
07:12a special speaker,
07:14Todd Valines, who is a
07:15professor of medicine
07:17and vice, chief of clinical
07:19services in the division of
07:20cardiology
07:21at the University of Virginia.
07:25And just to tell you
07:26a little bit about Todd,
07:27so he's, as I said
07:29a Professor of Medicine. He
07:31got his undergraduate degree,
07:33at West Point. He got
07:34his MD degree at the
07:36University of Oklahoma College of
07:37Medicine.
07:39He did his internal medicine
07:40training at Walter Reed Army
07:42Medical Center, and then he
07:44did a fellowship
07:45at the National Capital Consortium,
07:49at Walter Reed National Naval
07:50Medical Center.
07:52He was initially appointed on
07:54the faculty,
07:55the uniformed services,
07:57University
07:58of Health Sciences in Bethesda,
08:02and then he served on
08:03the US Army Command
08:06and General Staff College and
08:07he was a consultant to
08:08the US Army General Surgeon
08:12General and was a leader
08:13in Army Cardiology.
08:17He joined, the faculty at
08:18the University of Virginia in
08:20twenty nineteen
08:21and he's now the Chair
08:23of Clinical Operations,
08:26at that institution,
08:30But importantly
08:32were his contributions
08:33in the sign in the
08:34in the field of medicine.
08:37He is the past president
08:38of the Society of Cardiovascular
08:40CT,
08:41the immediate past editor in
08:43chief of the Journal of
08:45Cardiovascular
08:46CT.
08:47And in two thousand twenty
08:48one, he received the highest
08:50award, the gold medal award,
08:52from the Society of Cardiac
08:54CT.
08:56He also served as chair
08:58for both
08:59the ACC Imaging Section and
09:01the Federal Cardiology Section,
09:04past chair of the Certification
09:06Board of Cardiovascular
09:07CT,
09:08past president of the American
09:10Heart Association,
09:11Washington DC chapter, and he
09:13has innumerable
09:14military and academic awards and
09:17honors, a very impressive
09:19CV. He's published
09:20over three hundred publications,
09:22and he's really an international
09:24leader in cardiac CT. So
09:26it's really a pleasure to
09:27have him here and join
09:28me in welcoming Todd as
09:30our ground round speaker.
09:37Wow. You know, when when
09:39people come up and say
09:40that was an overly kind
09:41introduction, this really was
09:43an overly kind introduction. And
09:45and I wanna first start
09:46by,
09:48just, thanking all of you
09:49for this opportunity. This is
09:50truly an honor of a
09:51lifetime because
09:53to to not only, come
09:54to what I consider hallowed
09:55ground in the imaging world.
09:57I mean and and so
09:58for our residents, it's really
09:59exciting to have the trainees,
10:01for our residents, students, postdocs.
10:04You know, this is hallowed
10:05ground in imaging, and and
10:07it it is something that,
10:08you know, know, to have
10:08the opportunity to train with
10:09people like doctor Sanusis and
10:11the entire,
10:12group,
10:13and to hear the diverse
10:16areas of research that you're
10:17doing, which are truly groundbreaking
10:19and have changed practice for
10:20many of us who do
10:22multimodality imaging. So thank you
10:23for this opportunity, doctor Sanusas,
10:25and to see the lab,
10:27and the work and for
10:28doctor Velasquez.
10:30I I am just deeply
10:31honored, to and today, I'm
10:33gonna talk to you mainly
10:33about coronary CTA and and
10:36many
10:37of the, I I would
10:38say,
10:39emerging topics, but I think
10:41they're here in in in
10:42in the forefront today related
10:44to AI. And you heard
10:45all morning
10:46there rarely a talk that
10:47didn't talk about the potential
10:49or promise
10:50or or even the current
10:50use of AI in your
10:51research.
10:53These are just disclosures. I
10:54do some clinical trial leadership,
10:56that involves CT,
10:58acquisition, and so just making
10:59sure that those
11:01are appropriate, acquisitions. And I'll
11:02point out that while I'm
11:03talking about coronary CT,
11:06that,
11:08actually, we I guess we
11:12I think this is the
11:13old slide set, but,
11:15that's okay. But I I
11:16had some slides that I
11:17wanted to just also point
11:18out and thank,
11:20Renee. I had the opportunity
11:22to to have dinner with,
11:24doc doctor Barrett's,
11:26spouse, and and and I
11:27also,
11:29many years ago, got the
11:30chance to meet,
11:32doctor Barrett's
11:34daughter-in-law who was training in
11:35OB GYN at Walter Reed,
11:37and she just happened to
11:37be on my cardiology service.
11:39So, and I pulled out
11:41at that time, and I
11:41had a slide that just
11:42showed a picture of doctor
11:44Zaritz's book,
11:46which, I, like so many
11:48of you, that practice nuclear
11:49cardiology read.
11:50And so, you know, thank
11:52you for being here today
11:53and, and such a pleasure
11:55to have dinner and and
11:56and to meet you.
11:58So why don't we, we'll
11:59just go ahead and,
12:01get started.
12:02Real real quickly,
12:04is the
12:06do we have the other
12:07slide set?
12:10That's okay. We'll we'll roll
12:11with this.
12:14So what I'm gonna talk
12:15about originally is where we're
12:16at in corn in with
12:17with coronary CT in twenty
12:19twenty five.
12:20And so I think many
12:21of you are used to
12:22seeing the clinical indications for
12:24coronary CT. This is a
12:25noninvasive angiogram.
12:27You know, I was very
12:29I I you know, I
12:29was talking yesterday with doctor
12:30Velasquez about how I got
12:31interested in this technology and
12:33this thought that if you
12:34think back to when we
12:35were training,
12:36you know, invasive angiography, we
12:38were all doing functional testing
12:41with this probably overly singular
12:43focus on epicardial coronary disease.
12:45Right? We were focused on
12:47detecting what I show you
12:48here. Right? Severe epicardial disease.
12:49And we've learned there's so
12:50much more physiology that's important
12:52from functional imaging. We've learned
12:54about the microvasculature,
12:56INOCA, ANOCA, but this was
12:57really the goal. And the
12:59thought that you could do
12:59this noninvasively,
13:01accurately with a beating heart
13:03was quite,
13:04quite a novel idea back
13:06in the early two thousands.
13:07And having grown up doing
13:08coronary calcium,
13:09research,
13:11that was, you know, to
13:12be able to do angiography
13:13with CT was quite based.
13:14So most of you, when
13:15we're talking about the clinical
13:16utilization, obviously, in symptomatic patients,
13:19identifying patients like this is
13:20a super helpful role, this
13:22ability to do noninvasive angiography.
13:24But I think what got
13:25me most excited about
13:27coronary CT was this idea
13:29that not only can we
13:30identify
13:31high risk epicardial coronary disease,
13:33perhaps better select patients who
13:35may benefit from revascularization
13:37or who need or don't
13:38need the cath lab,
13:40but But this idea that
13:40we can see three-dimensional whole
13:42heart atherosclerosis
13:43in a noninvasive test and
13:45to use that to then
13:47change
13:48our preventive
13:49approach
13:50and to harness this data
13:52and to change management.
13:53Because, you know, this was
13:54something that, we had seen
13:55this concept of seeing atherosclerosis
13:57and treating atherosclerosis with coronary
13:59calcium score. And so if
14:01you look, you know, at
14:01current state, look at current
14:03guidelines, we know that coronary
14:04CT has come a long
14:05way. When you compare this
14:07to prior chest pain guidelines,
14:08it now is a recommended
14:10first line test in patients
14:12who are,
14:13who are without known coronary
14:14disease. And so this is
14:16something that, you know, I
14:17think when we think about
14:18test selection, where the really
14:20the the foundational role of
14:21CT is in patients who
14:22have no known coronary disease
14:24and the guideline writers have
14:25talked about when you might
14:26choose coronary CT and when
14:28you might be more more
14:30more preferred to choose functional
14:31testing. And so I think
14:32this is data that you're
14:33all aware of and this
14:34is true for both acute
14:35and chronic chest pain.
14:37Now the guideline writers, much
14:38what I just mentioned about
14:40this concept of atherosclerosis,
14:42I think, you know, really
14:43changed the phenotypic description of
14:45who has coronary disease. And
14:46so if you think traditionally,
14:48if you looked at clinical
14:49trial enrollment or if you
14:50looked at,
14:51how we define people with
14:53established coronary disease clinically,
14:55we typically use stenosis
14:57to define it. It was
14:59based on how much stenosis
15:00you had. Right? And so
15:02typically, you had to have
15:02at least a fifty percent
15:04stenosis or some threshold, seventy
15:05percent in non lane left
15:07main vessels. You have to
15:08have had an event
15:09or a revascularization
15:11procedure to be defined. And
15:12I think the guideline writers
15:13for the first time said
15:14that atherosclerosis
15:16identified
15:17on CT
15:19is coronary
15:20artery disease
15:22and that we should use
15:23this information to pair this
15:25commensurally,
15:26the intensity of prevention to
15:27disease severity. And so our
15:29European colleagues have have, similar
15:31guidelines. This is now, what
15:33I just showed you, what
15:33we're doing in the US
15:34with coronary CT is actually,
15:37now being done
15:38internationally. And in fact, you
15:40know, these these are actually
15:41just recent guidelines. This has
15:42been the case for now
15:43for many years. Now these
15:45guidelines are based on prospective
15:47randomized clinical trials to have
15:48a class one recommendation and
15:50guidelines,
15:51and they're based on data
15:52showing that,
15:53you know, we know it's
15:54not the tests that essentially
15:55change the outcome. It's how
15:56you utilize this information. And
15:58so if you look at,
15:59for example, in Scott Hart,
16:01that when you paired
16:03atherosclerosis imaging with an increased
16:05use of preventive therapies, that
16:06you saw about a forty
16:07percent reduction in MI. And
16:09so this concept that, you
16:10know, we're not making people
16:12necessarily live longer because we're
16:14putting in more stents or
16:16necessarily doing more bypass surgery
16:17with CT, but it's in
16:18fact this identification
16:20of nonobstructive disease paired with
16:22preventive therapies. And I think
16:23this is the concept that
16:24we're most excited about. And
16:25so just to kinda talk
16:26about where we're at in
16:27twenty twenty five and how
16:29AI software is potentially changing
16:31practice, I think it's important
16:33to understand the strengths but
16:34also the limitations of anatomic
16:36imaging. And I think no
16:37there's no place in the
16:38world that understands that better
16:39than the people in this
16:40room.
16:40But we know the real
16:41value of coronary CT is
16:43its very high sensitivity
16:44for detecting both angiographically
16:47significant coronary disease, so using
16:48a fifty percent stenosis threshold,
16:52as well as detecting functionally
16:53significant,
16:55CAD based on an abnormal
16:56invasive FFR.
16:58And so we know in
16:59the in this scenario that
17:00there's, you know, stenosis less
17:02than fifty percent. You're very
17:03unlikely to have hemodynamically significant
17:06disease if used in FFR
17:07standard,
17:08or you're unlikely
17:09to to
17:11be missing significant angiographic disease.
17:13Now on the con on
17:14the converse, right, we know
17:15that its specificity
17:16as a static image is,
17:18about the same as other
17:19tests, modest,
17:21and that when we see
17:22intermediate stenosis
17:23on CT, that it's probably
17:24a flip of a coin
17:25as to whether that is
17:26a functionally significant lesion. And
17:27so that's certainly the the
17:29the downside of CT.
17:31So how is CT today
17:32changing who how we treat?
17:34And I'm gonna show you
17:35some recent data,
17:37and then kind of with
17:37an eye forward to how
17:39perhaps the field is is
17:41is changing. Now we know
17:42that historically
17:43we have done a relatively
17:45average job at selecting patients
17:48for invasive angiography.
17:51When and and I say
17:52historically, if you look at
17:53large scale, NCDR, these are
17:55ACC run, cath registries if
17:57you look at the USVA.
17:59If you take patients without
18:00established coronary disease, who most
18:02of which have had either
18:03no testing,
18:04but most of them have
18:05had functional testing. It's about
18:07half of patients end up
18:08leaving the cath lab with
18:10a with a a diagnosis
18:11of of of minimal to
18:12nonobstructive coronary disease.
18:15And so can we do
18:16better? And I say that
18:17because, you know, if you
18:18look at my cath lab
18:19at the University of Virginia,
18:21my interventionalists
18:22like doing things and fixing
18:23things and doing normal cast
18:25is not a you know,
18:26you know, most people aren't
18:28super thrilled about that, but
18:29it's really more about efficiency
18:30of care and resource utilization.
18:32You know, I think most
18:33of us would argue that
18:34with our noninvasive tools, the
18:35people in this room, we
18:36we can phenotype and pretty
18:38accurately diagnose people noninvasively.
18:40So people who go to
18:41the cath lab are people
18:41who actually
18:43are likely to need revascularization
18:45or they need something we
18:46can't do noninvasively.
18:47And so the question is,
18:48can we do better? And
18:50and if you look at
18:51data,
18:52you know, we know that
18:53if you use CT, you're
18:54much less likely to have
18:55a normal cath when you
18:56leave the cath lab. And
18:58this is even true in
18:59recent patients when we look
19:00at studies like ischemia. These
19:01were patients with moderate to
19:02severe ischemia based on sight
19:04reads. And when you did
19:05coronary CT in about eighty
19:06plus percent of these patients,
19:08one in five of them
19:09did not have significant angiographic
19:11disease. This is not to
19:12say that they didn't have
19:12things like microvascular disease or
19:14other things, but they probably
19:15didn't need to go to
19:16the cath lab.
19:17And we could probably use
19:18those resources perhaps a little
19:20more,
19:20appropriately.
19:21And so this is a
19:22relatively recent study. This this
19:24was a study that was
19:25started before the guidelines, but
19:27it came out after the
19:28guidelines.
19:29This was, a sponsored study
19:31by by one of the
19:32vendors in the field, but
19:33they, were comparing
19:35a CT guided strategy with,
19:38optional
19:39FFR CT. I'm not gonna
19:40be talking much about that
19:41technology.
19:42About a third of patients
19:43got that if they had
19:44intermediate stenosis versus functional testing.
19:46This is a relatively low
19:47risk outpatient population.
19:49And again, what did they
19:50find? Well, no difference in
19:52death or MI,
19:53big reduction, about a seventy
19:55percent reduction in unnecessary casts.
19:57And so this concept of
19:58can we be more efficient
19:59with our resources,
20:01in a in a scenario
20:03where we're having increasing demands
20:05for, spots in our testing
20:07procedures, our labs, and especially
20:09our our cath labs. And
20:10so this just shows it
20:10another way. In patients
20:12who had CT, less than
20:14thirty percent of them, left
20:16the cath lab without revascularization
20:18or a plan for CABG,
20:19and this was exactly opposite
20:21in patients who got usual
20:22care. About seventy percent of
20:23patients left the lab without
20:25revascularization.
20:26Sixty percent had no significant
20:28angiographic coronary disease at all.
20:30And again this was just,
20:31essentially in many ways observational,
20:34in the sense that people
20:35sent to the cath lab
20:36is left up to local
20:37site investigators.
20:39So what we're seeing now
20:40in a current guideline era
20:41is that,
20:43utilization
20:44of coronary CT has significantly
20:46increased. This is actually some
20:47older data from just, about
20:49three months ago. And if
20:50you just look at these
20:51this kind of exponential increase
20:53in our,
20:54coronary CT lab and this
20:56is actually now this is
20:57actually not even, the full
20:59year for twenty twenty five.
21:01We've seen that it's now
21:02at UVA become the dominant
21:03testing
21:04modality for ischemic heart disease,
21:06at our institution.
21:08And while all of our,
21:09imaging modalities are growing, we've
21:11really seen this increased
21:13utilization of coronary CT. And
21:14so this has really created
21:15some unique problems, and this
21:17will dovetail into what we
21:18talked about using software. When
21:19I say unique problems, it's
21:20how do we schedule these
21:21patients?
21:22How do we get these
21:23people in and out of
21:24the scanner
21:25as quickly as we can
21:26so we utilize these slots
21:27appropriately? How do you know,
21:29we're really traveling to drill
21:30down and look at every
21:31part of
21:32the patient experience,
21:34premedication,
21:35where we start IVs, etcetera
21:37and so forth. And so
21:38we've now built the use
21:40of coronary CT into our
21:42this is actually called Agila
21:43MD. You probably have this
21:44embedded within your
21:46EHR and Epic. And so
21:47this is our observation pathway
21:49for people with intermediate troponins.
21:51And this is now,
21:52you know, the recommended testing
21:54strategy for, and we arbitrarily
21:56set this at anyone under
21:57eighty years old without known
21:58coronary disease.
22:00And so this is, kind
22:02of the approach. There's an
22:03order set that's built into
22:04this, and then what we've
22:05provided are recommendations
22:07for management
22:08as well as follow-up care
22:09and when to consult patients
22:11based on, in this case,
22:12a relatively simplistic,
22:14reporting tool called CADRADS.
22:16But this is something that
22:17that I I'm not I'll
22:18I'll spend a little bit
22:19of time about.
22:20But this is now,
22:21you know, again, this this
22:22challenge of how do we
22:23get these patients through red
22:25quickly in in a in
22:26a scanner
22:27scanned in an efficient way
22:29is really a challenge. I'm
22:30gonna talk a little bit
22:31about the current reporting of
22:32CT, and this is, I
22:34think, again, setting the stage
22:35of where we are in
22:36twenty twenty five
22:38because there are significant changes
22:39that are about to occur
22:41in,
22:42some software,
22:44that may, be available. And
22:46and the question is then
22:47how how does this help
22:48us? And and my question
22:49for you,
22:50and I wanna have you
22:51think about this critically, is
22:53does this help us? Do
22:54we need these, advanced AI
22:55tools?
22:57Well, here's the current state.
22:59And so the current state,
23:00this is I mentioned CADRADS.
23:01This is a document. It's
23:02a multisocietal document led by
23:04the Society of Cardiovascular CT,
23:06and the whole goal,
23:07is just to improve
23:09consistency. If you look at
23:10some of the older CT
23:11reports or studies like PROMIS,
23:14highly variable reporting.
23:16No consistency or minimal consistency
23:18in what people were calling
23:19mild, moderate, severe. There was
23:21very, very scattered recommendations or
23:23many times no recommendations
23:25as to what to do
23:26with this information.
23:28And the thought was to
23:29standardize this. And so I
23:30think you're all familiar with
23:31this. This is the current
23:32scheme that people have their
23:34coronary CT reported,
23:36based on worse stenosis.
23:38And the initial version, CADRADS
23:40one point o, this is
23:41all that there was in
23:42the sense that there was
23:44a comment about high risk
23:45plaque.
23:46But what the second version
23:47that I just showed you
23:48did is it's it mandated
23:50that all patients who undergo
23:51coronary CT have a an
23:54assessment, not just of worse
23:55stenosis,
23:57but of the overall plaque
23:58burden.
23:59How much plaque do you
24:01have? And the reason for
24:02this is that and we're
24:03not gonna go through all
24:04these studies, but we've seen
24:05consistent
24:06consistent results, whether it be
24:07large registries or prospective randomized
24:09clinical trials,
24:10that in addition to stenosis
24:12stenosis is prognostically very important.
24:15But in addition to that,
24:16probably the strongest predictor of
24:18risk
24:19is the overall extent of
24:20atherosclerosis.
24:22And our previous versions of
24:23CADREDS did not include that.
24:25It could say you could
24:26be CADREDS one and have
24:27a single plaque, and you
24:28could be CADREDS one and
24:29have thirty plaques,
24:30and we called them the
24:31same.
24:32And so now in CADREDS,
24:34two point o, you can
24:35see that there is some
24:36assessment of overall plaque burden.
24:38And the question is, how
24:39do we measure that?
24:40Right. Well, here are some
24:42options you can do. If
24:43you do a calcium score
24:44prior to your coronary CTs,
24:45which we've done that international
24:47surveys, about seventy percent of
24:49labs do routine calcium scoring.
24:51Some do it selectively,
24:52avoiding it in younger patients.
24:54Some do it in everybody
24:56with the thought being that
24:57it provides additional,
24:59prognostic information perhaps that people
25:01are used to seeing or
25:02maybe understand.
25:05It may change how you
25:06scan the patient in scenarios
25:07where you have very dense
25:08extensive coronary calcium.
25:10But suffice to say, if
25:11you do that, this is
25:12one recommendation.
25:13These are pretty standardized categorizations
25:16of mild, moderate, severe, or
25:17extensive coronary calcium.
25:19You can simply count the
25:21number of segments with atherosclerosis.
25:23There is prognostic data supporting
25:24this role,
25:26or you can just visually
25:27estimate
25:28the number of vessels and
25:29the extent within those vessels
25:31and do a virtual score.
25:33And so you can see
25:34some of the challenge and
25:36challenges in this, particularly if
25:37you're not doing calcium scoring,
25:39which we know calcium scoring
25:40has reasonable reproducibility.
25:42And so this is the
25:43the current state today.
25:45In addition to that,
25:47we report the presence
25:49of what's called high risk
25:50plaque. I'm gonna come back
25:51to this topic and talk
25:52when we talk about
25:54AI and plaque quantification.
25:56And currently, high risk plaque
25:57is defined as any plaque
25:59with at least two of
26:00these features, and I've just
26:01shown you visually what these
26:02look like.
26:04And you can see here,
26:05you you know, two classic
26:07positive remodeling, low attenuation plaque
26:09if you've got plaque with
26:10Hounsfield units of less than
26:12thirty, spotty calcification, or napkin
26:14ring signs. So this is
26:15the reporting
26:16today. This is,
26:18you know, about three years
26:19old.
26:20And what is this based
26:21on?
26:22And I'm gonna I'm gonna
26:23kinda, you know, spend the
26:24rest of this talk looking
26:25behind the scenes and challenging
26:27a little bit of what
26:28we do,
26:29and challenging a little bit
26:30about what some people think
26:31we ought to do.
26:33This is why we report
26:34high risk plaque. This was
26:35a seminal paper from Sudeki
26:36Motoyama's group. She took a
26:38little over a thousand patients.
26:39They all had nonobstructive
26:41coronary artery disease on a
26:42coronary CT,
26:44and and she followed them
26:45prospectively.
26:47And what they noticed is
26:48that,
26:49and and these are all
26:50patients who were followed and
26:51had and if they had
26:52events, they had data on
26:53culprit lesions.
26:55And and they looked at
26:56these patients and said, you
26:57know, what was predictive
26:59beyond stenosis?
27:00And it turns out these
27:02two feature plaques that were
27:03low attenuation
27:04and positive remodeling, only about
27:06five percent of patients had
27:07these.
27:08But you had a twenty
27:09fold increased risk of ACS
27:10if you had these.
27:12It didn't always predict the
27:14culprit lesion.
27:15So patients who had these
27:17often developed new high risk
27:18plaques.
27:19And, in fact, some of
27:20these plaques over time regressed.
27:22But what was shown is
27:22that this was identification of
27:25at least a high risk
27:26patient who has propensity to
27:28develop high risk plaques. And
27:29this has been subsequently shown
27:30in in NHLBI funded studies
27:32such as PROMISE,
27:34that I think doctor Velasquez
27:35was involved with and and
27:37and and and and probably
27:38several others in this room,
27:40again showing that in addition
27:41to traditional risk factors, calcium
27:43scorching stenosis, that the presence
27:45of high risk plaque in
27:47a binary way
27:48was associated with increased,
27:50events.
27:51This is an example of
27:52a patient that stand at
27:53our institution. This is a
27:54sixty year old man, pretty
27:55nonanginal chest pain, was a
27:56smoker,
27:57had zero coronary calcium,
28:00had a plaque that looked
28:01like this. I don't know
28:02if you can appreciate this.
28:03This degree of stenosis was
28:04not impressive,
28:05but this person had a
28:07lot of non calcified plaque.
28:09There was expansile remodeling or
28:11positive remodeling, low attenuation,
28:13and this was a person
28:13who came back about five
28:14months later with an anterior
28:16infarct right at that particular
28:18location.
28:19And he, unfortunately, elected because
28:21his calcium score was zero,
28:23and he said, I'm not
28:24gonna take a statin. I'm
28:25gonna keep smoking. But this
28:26is just an obviously extreme
28:28example. We know the risks
28:29of of these types of
28:30events are low in people
28:31with calcium scores of zero,
28:33but plaque biology,
28:35does matter.
28:36And so I want to
28:37spend a little bit of
28:38time as we start to
28:39talk about machine learning is
28:40what does this mean from
28:41a pathological
28:42standpoint? You know, when I
28:44would go out and talk
28:44about high risk plaques, people
28:46my interventional
28:47colleagues,
28:48Greg Stone would say, what
28:49are you guys in CT
28:50talking about?
28:51This is high risk plaque,
28:53and we know this from
28:54an in invasive cardiology.
28:56And you may know this,
28:57many of you, from histology.
28:59And no one talks about
29:00spotty calcification
29:02and low attenuation and positive
29:04remodeling
29:05in those two fields.
29:07So this is data from
29:08the prospect two trial, and
29:10these were patients who came
29:11in with ACS. They got
29:13three vessel IVUS with virtual
29:14histology,
29:15and they then followed these
29:17patients prospectively
29:18to look at
29:20what happened to them and
29:21where
29:22the event, future ACS, if
29:24it did occur, where did
29:25it occur? And what they
29:26found is that, of course,
29:27if you have a documented
29:29thin cap fibroarthroma,
29:32then you're at higher risk.
29:33We know that from
29:34years of
29:36If you have stenosis, right,
29:38stenosis is a high risk
29:39feature. Minimal luminal area of
29:40less than four millimeters squared
29:41in a proximal to mid
29:42coronary segment.
29:44If you have plaque burden
29:45so, again, this concept that
29:47on a cross sectional image
29:48of more than seventy percent
29:49of that area is occupied
29:51by plaque,
29:52kind of the invasive analogy
29:54analogies
29:55analogy to positive remodeling.
29:58And if you had all
29:58of these things, you can
29:59see your more than eighteen
30:00fold
30:01higher rate of having ACS
30:03irrespective of clinical risk factors
30:05or your prior angiogram.
30:07And, again, it didn't always
30:09predict the culprit lesion,
30:11but it predicted patients more
30:12likely to develop sometimes new
30:14high risk plaque. And so
30:15when interventionalists talk about high
30:17risk plaque, this is what
30:18they're referring to. And so
30:19the question is, what are
30:20we detecting on CT compared
30:22to this?
30:23And this is just a
30:24summary slide showing from if
30:26you look at, you know,
30:27these these these multiple,
30:29really,
30:31you know, seminal trials that
30:32taught us about plaque biology
30:34assessed noninvasively,
30:36of the hazard ratios associated
30:37with thin cap fiber atherosclerosis,
30:40plaque burden, high plaque burden
30:42being the most predictive,
30:44if you look at MLAs
30:45less than four, and then
30:46another technology called near infrared
30:48spectroscopy
30:49that identifies
30:50these lipid these these highly
30:53highly lipidic plaques or high
30:54lipid burden plaques.
30:56These are what most interventionists
30:58consider high risk plaque.
31:00So what are we seeing
31:01on CT? Well, if we
31:02go back, to our roots
31:03here, these are there are
31:04multiple small studies. These are
31:06hard studies to do, by
31:06the way. These are not
31:07thousands of patients. But this
31:09is a small study. I
31:10could show you four other
31:11similar studies, but they took
31:13a small number of patients
31:14who were getting intravascular imaging.
31:16And this time, they used
31:17even a better technology than
31:18virtual eyes. They used optical
31:19coherence tomography,
31:21and they which is really
31:22the the single modality that
31:24can identify truly at the
31:25highest
31:26accuracy thin capped fibroarthromas.
31:28And what they found is
31:29that true thin capped fibroarthromas,
31:31the most risky plaques
31:33from a biological standpoint,
31:36eighty percent of them had
31:38two adverse features.
31:40So these high high risk
31:41plaques are that we're seeing,
31:42we're describing,
31:43are more likely to be
31:45than capped by breath rooms.
31:47We cannot image down to
31:49the resolution to see cap
31:51thickness on CT, not even
31:52with photon counting CT scanners.
31:55And which two was it?
31:56Well, again, it was these
31:58two positive remodeling
32:00and low attenuation plaque. So
32:02those features were more than
32:03ten to fifteen fold more
32:05likely to be a thick
32:06cap fibroarthromas.
32:07So when someone says, what
32:08does this mean, this high
32:09risk plaque? You're scaring people.
32:12Do I need to go
32:12in and stent it? Well,
32:13the answer, of course, is
32:14we hope we're not scaring
32:14people, and, no, you shouldn't
32:16do preemptive stenting. There are
32:17trials that are studying that.
32:19That's really interesting.
32:20Greg Stone, of course, wants
32:22to do that in everybody,
32:23but you gotta have science
32:24for that. But what it
32:25does mean is that perhaps
32:26you should consider this risk
32:28above and beyond the stenosis
32:30severity and the calcium score
32:31like the patient I just
32:32showed you.
32:34So now what about lipid
32:35rich plaque and where are
32:36we gonna use software to
32:38perhaps help us?
32:39This is a study where,
32:41a software was used,
32:43to quantify
32:45plaque
32:45burden. So it can quantify
32:47the volume of low attenuation
32:49plaque. And it said, how
32:50do these plaques that have
32:51a lot of low attenuation
32:52plaque compare to lipid
32:54content
32:55using near infrared spectroscopy?
32:58And so this is,
32:59you know, work using, again,
33:00one of these software. And
33:01what they found was there
33:02were that plaques who had
33:04more than two point three
33:05cubic millimeters of low density
33:07plaque or low attenuation plaque
33:09were over ninety percent
33:11sensitive and specific for predicting
33:14high risk plaque defined by
33:15NEARS.
33:16And so, again, this concept
33:17that when we see low
33:18attenuation plaque, these are not
33:20only more likely to be
33:21thin capped bibryotheromas,
33:22they're also more likely to
33:23be high lipid content plaques.
33:26And so I think this
33:26is helpful as we think
33:28about how we manage patients
33:30in the current, era. So
33:31I think this is just
33:32a slide which reminds us
33:34that while we can't see
33:35plaque thickness,
33:36and we're reporting plaque burden
33:38relatively semi quantitatively,
33:40that
33:42current CT does a decent
33:44job at identifying
33:46plaques that might
33:48portend an increased patient level
33:49risk and maybe change management.
33:51The concept that plaques that
33:52look differently
33:53do behave differently. In fact,
33:55what work that we have
33:55done has shown that in
33:57fact on the end spectrum,
33:59that calcification
34:00is in fact a healing
34:01process, and in many ways,
34:02a plaque stabilization
34:04process. So if you look
34:05at what are called one
34:06k plaques, plaques that have
34:07more than a thousand Hounsfield
34:09units,
34:10almost never rupture and cause
34:11an ACS event. In fact,
34:12they're almost the lowest risk
34:14plaque that we have. Now
34:15they can certainly be flow
34:16limiting,
34:17particularly in the proximal vessels,
34:19but they're very unlikely to
34:20cause an ACS event from
34:22plaque rupture.
34:23And so this is kind
34:24of, I think, what we've
34:25learned. Now what's the dirty
34:26little secret about all of
34:27this?
34:28Well,
34:29the challenge is that if
34:30I look at a CT
34:31and my colleague looks at
34:32a CT and I say
34:33that's high risk plaque,
34:36We may agree, but there's
34:37a good chance that we
34:38won't. Meaning that we know
34:39from very good studies that
34:40there's only moderate interobserver
34:42agreement
34:43for high risk plaque features.
34:45And this has a lot
34:46to do with a lot
34:47of variables. A lot of
34:48it's due to image quality.
34:50Right? We know there's highly
34:51variable image quality in clinical
34:53CTA. And so this is
34:55something that perhaps perhaps we
34:56can get better at. And
34:57and the other question, of
34:58course, is can software help
34:59us? We'll see.
35:01But even in research settings
35:02and experienced labs, the inner
35:04observer agreement for
35:06identifying high risk plaque is
35:08not great. In fact, this
35:09is a group in Europe
35:10who just recently published. This
35:12is this quantitative coronary imaging
35:14group who's trying to bring
35:16some standardization
35:17to how we quantify coronary
35:19disease in an era
35:21of an increasing use of
35:23software tools. And what they
35:25said is that essentially for
35:26high risk black, these are
35:27really the only two features
35:28you should be reporting, positive
35:30remodeling, low attenuation, those two
35:31that seem to be the
35:32most predictive
35:34because very low levels of
35:35agreement,
35:36if you look at the
35:37published literature for identifying spotty
35:39calcification and napkin ring sign.
35:40I think most of us
35:41who read a fair bit
35:42of CT would tell you
35:43that's the case for those
35:44two features. So spotty calcium,
35:46maybe not so much, but
35:47napkin ring sign is is
35:48is it's it's it's it's
35:51not not an easy one
35:52to see consistently.
35:53So that's kinda where we're
35:54at in the field. And
35:55what I want us now
35:56looking is where we're going
35:57in the field and moving
35:58beyond twenty twenty five. And
36:00one is that there's tremendous
36:01interest in moving away from
36:03use defining
36:04disease by stenosis.
36:06And so this was, just
36:08a recent Lancet Commission. The
36:10front cover of
36:12Lancet from that month said
36:13that they thought that by
36:14by refocusing
36:16how we address corn disease
36:17worldwide,
36:18that we could they could
36:19save over eight million lives,
36:21per year.
36:22We'll see if that comes
36:23to fruition, but this is,
36:25a challenge, and I advise
36:26you to kinda read through
36:27this, that as a field,
36:29we need to think about
36:30plaque burden
36:31and atherosclerotic burden,
36:33in how we, address patients
36:35from a preventive standpoint.
36:37And I think one of
36:38the two I one of
36:39the talks that you just
36:40heard today about population
36:42opportunistic screening with non gated
36:44chest CT is just one
36:46example of that, is using
36:47data that we have for
36:49opportunistic
36:49screening.
36:50So this is, if you
36:51go back in time when
36:52I was really just about
36:54a year into doing, you
36:55you know, you know, CT
36:56as an as an attending,
36:58you know, this was the
36:59the call, to all of
37:00us. I got very nervous
37:01when I saw the front
37:02cover because there's absolutely no
37:03data in two thousand five
37:05that CT could prevent heart
37:06attacks.
37:07We were doing CT at,
37:09you know, decent image quality,
37:11I would say, but scanner
37:13technology was not great. You
37:14can even see if you
37:15have the front cover of
37:16this image.
37:17We were doing a very
37:18high radiation dose,
37:20and we didn't have data.
37:22So the question is, can
37:23we, you know, you know,
37:24where where have we come
37:25in since since then? And
37:26you saw the guidelines. You
37:27saw Scott Hart. But we
37:28you know, the question is
37:29where are we going? Well,
37:31you know, this is data
37:32that goes way back, and
37:32I'm just gonna remind us
37:33that stenosis still matters. And
37:35we've known this. This is
37:36a very large registry, almost
37:37twenty four thousand patients. The
37:39more vessels with stenosis that
37:40you have, remember stenosis is
37:41a high risk feature.
37:43But we have been trying
37:44to quantify what we're seeing,
37:46and you saw the Cad
37:47RADS two point o attempts
37:48at quantifying plaque burden.
37:51How do we do that?
37:52Well, you know, this is
37:53very old data that simply
37:54just counting the number of
37:55segments with disease maybe is
37:57prognostically important. Segment involvement score.
37:59This is another score. I'll
38:00go through these pretty quickly.
38:01This is a Leiden score.
38:02We were really proud of
38:03ourselves within the confirmed registry
38:05that we came up with
38:05a smarter score that no
38:07one uses.
38:08And so no one uses
38:10this today. Right? We said,
38:11okay. Well, let's look at
38:12the prognostic
38:13weight of the proximal location,
38:16the type of plaque,
38:17the stenosis, and put all
38:19of this in a score.
38:20And then we added to
38:21this clinical factors, called it
38:22the confirmed risk score. No
38:24one uses this stuff.
38:26And you know it's and
38:27we said this is even
38:28better than the segment involvement
38:29score, better than other scores.
38:30And the problem is no
38:31one's doing because they're they're
38:32clunky to
38:34do, and it's still fairly
38:35quite it's fair still fairly
38:37subjective.
38:38And so I think what
38:39the field has realized is
38:40that, you know, there's potential
38:42using
38:43AI tools to truly quantify
38:44three-dimensional
38:45atherosclerosis. And this is really
38:47one of the first papers
38:47to show it show this.
38:49This is data from Scott
38:50Hart. Remember, these patients were
38:51randomized to either coronary CT
38:53or standard of care.
38:54And in the CT patients,
38:56about seventeen hundred patients from
38:58Scotland who were symptomatic,
39:00they quantified
39:01plaque volumes,
39:02and they said, what is
39:03most predictive of events? Now
39:04I'll say this is a
39:05relatively low risk group,
39:07but what they showed is
39:08that the volume of low
39:10attenuation plaque was the strongest
39:12predictor of incident MI. It
39:13was better than calcium scoring.
39:15It was better than stenosis.
39:16It was better than clinical
39:17risk factors, and they showed
39:19some thresholds. For example, if
39:20you had more than four
39:21percent
39:22of your coronary,
39:24tree or your plaque burden,
39:27as low attenuation plaque. They
39:28saw, you know, this fivefold
39:30increase in myocardial infarction even
39:32when accounting for these other
39:33factors. And there have been
39:34numerous groups, and this is
39:35just from Dominique Day's group
39:37at Cedars Sinai, where they
39:38have gone and used multiple
39:40populations to train software
39:42and compare it to intravascular
39:45imaging for accuracy, for volumetric
39:47assessments,
39:48comparing it in some cases
39:49to,
39:51you know,
39:52large populations for outcomes,
39:54and even some even some
39:56small histology groups showing that
39:58software can do a reasonable
39:59job
40:00if you compare the volume
40:02of
40:03plaque to things such as
40:04intravascular
40:05ultrasound,
40:07or even manual assessments of
40:08plaque on CT. This can
40:10do this in this particular
40:11study. They could do this
40:12in just a few seconds,
40:13with some oversight as to,
40:16ensuring that the vessel was
40:17segmented appropriately.
40:19And in fact, in this
40:20in this analysis, they showed
40:21from Scott Hart that there,
40:22you know, plaque volumes above
40:24two hundred and thirty eight
40:25and a half millimeters cube,
40:26you saw this significant sevenfold
40:29increase in risk.
40:30And this was,
40:32you know, certainly a better
40:33predictor for outcomes than stenosis.
40:36And we don't currently do
40:37this. Right? We don't currently
40:39report this. And so this
40:41has kind of gotten people
40:42a lot really, really, kinda
40:44interested in this field that
40:45should we be quantifying
40:47more. And, you know, at
40:48the end of this talk,
40:49you're gonna say, well, you
40:50know, that all sounded pretty
40:51good. Why why is doctor
40:52Vollandz a little bit negative
40:53about this field? Well, I'm
40:54gonna show you kind of
40:55why,
40:56in the next few settings.
40:57There are now software that
40:58are available, commercially available,
41:01that have taken this approach
41:02and said, we can do
41:03this for you.
41:04And we can stage this
41:06much like we do
41:07oncology patients or cancer patients.
41:10You know, that how how
41:11what do these numbers mean
41:12if you're gonna quantify plaque
41:13volumes? And so they they've
41:14tried to give some,
41:16you know, guidance to clinicians.
41:18And this is just a
41:19very early stage approach of
41:21trying to stage how bad
41:22is my coronary disease.
41:24Right?
41:25And this is just one
41:26such approach. Total plaque volume,
41:28you can see these different
41:28cut points, or percent atheroma
41:30volume, which is how much
41:31of the vessel volume is
41:33made up of plaque
41:34as one approach. And this
41:35was really based on invasive
41:37angiography, by the way. They
41:38took patients who had one
41:40vessel, two vessel, and three
41:41vessel obstructive disease, and they
41:43quantified their plaque
41:45and said these cut points
41:46seem to work pretty well.
41:49And so this is how
41:50this was done. Probably not
41:51the ideal way of deriving
41:53this, and we can argue
41:54about some of these cut
41:55points. I just showed you
41:56a sevenfold increase in risk
41:57if you were below two
41:58fifty for the Scott Heart
41:59pryper. So may you know,
42:01is is is stage one
42:02mild plaque really mild?
42:04So this has been criticized
42:05for that reason. But they
42:06have shown
42:07in small populations and just
42:09published this week actually from
42:11from a a separate population
42:13that plaque volumes
42:15are a little bit better
42:17than calcium scoring,
42:19stenosis,
42:20and the standard things. And
42:21this is looking at,
42:23ten year event rates.
42:24According to this stage, and
42:26what you can see is
42:27that the ten year event
42:28rates were relatively predicted by
42:31this staging system that I
42:32just showed you,
42:34in a moment.
42:35Not something we can do.
42:36And so as I mentioned,
42:37there are vendors out there
42:38that will come into you
42:39and talk with you and
42:40try to sell you this
42:41technology
42:42where they say, send us
42:43a CT scan.
42:45We'll process it three dimensionally.
42:46We'll send you back a
42:47report that has total coronary
42:49plaque volumes
42:51to include the volumes of
42:52noncalcified plaque. Remember this this
42:54this riskier low density plaque,
42:57calcified plaque,
42:58percent atheroma volumes, and we'll
43:00even assess stenosis for you
43:01automatically.
43:03Right?
43:04And so we we can
43:04do this for you. We
43:05can make this easier for
43:06you. There's there's another shot
43:08of another vendor where they
43:09will, you know, again, process
43:11these scans and take this
43:12data from you, analyze them,
43:14and send them back to
43:15you, giving you these plaque
43:17analysis.
43:20And in this scenario, instead
43:21of staging your disease, this
43:23particular vendor says, well, I'm
43:25gonna tell you how you
43:26compare to other men and
43:27women of the same,
43:29you know, men and women
43:30of the same rough age
43:32and give you a percentile,
43:33much like we do with
43:34calcium score. So you would
43:35get a coronary CT and
43:37would come back and say
43:37you're at the eightieth percentile
43:40compared to other
43:41men your age, for example,
43:42or women your age. And
43:44so this is another approach
43:45that's being taken.
43:46And they have shown, of
43:48course, not surprisingly,
43:49that patients in the highest
43:50percentile have a significant
43:53increased risk when you look
43:54at large populations. And, again,
43:56this is based on not
43:57only Scott Heart data, but
43:58other, another large population from
44:00Europe.
44:01But here's the
44:03rub, is that number one
44:04is that that the field
44:05is currently lacking real standardization.
44:07And what I mean by
44:08that is I think all
44:09of us as physicians like
44:10measuring what we see. There's
44:12so many areas in science
44:13where that actually does really
44:15improve patient care.
44:16But if you look at
44:17this is a statement from
44:18the SECT again trying to
44:20help with standardization,
44:21and they looked across multiple
44:23different software.
44:24And you can see that
44:25just based on Hounsfield unit
44:27thresholds, there's high variability
44:29in what people call calcified
44:31plaque, noncalcified
44:33plaque, and low attenuation plaque.
44:34And in fact, they ran
44:35a single scan with seven
44:36different software, and what they
44:38found were highly discrepant results.
44:40And and this is just
44:41an example of some of
44:42these different software. In fact,
44:43these results vary by more
44:45than a hundred and fifty
44:46percent with some plaque types.
44:49Why is this important for
44:50you to know this? Because
44:52surprising to many,
44:53starting on January first,
44:55Medicare has said we are
44:57paying for this.
44:58And this this is you
44:59can actually see this is
45:00a a an LCD, a
45:01local coverage decision
45:03from Medicare that says in
45:05patients
45:06who are symptomatic,
45:07who don't have known coronary
45:09disease,
45:10who obviously are eligible for
45:11coronary CT and don't have
45:13severe stenosis.
45:15Right? This is stenosis less
45:17than seventy percent.
45:19Then in those patients, you
45:20can do this plaque analysis,
45:22and they will pay nine
45:23hundred and fifty dollars and
45:25fifty cents for this.
45:26So that's the current payment.
45:29And they say, where can
45:30you not do this? Well,
45:31you shouldn't do this in
45:32screening cases.
45:34You shouldn't do this in
45:35people who have had a
45:35prior CT who are just
45:37doing it for
45:38disease progression.
45:41You know, you shouldn't do
45:42it if you're taking them,
45:43you know, to the cath
45:44lab for whatever reason. So
45:45these are kind of the
45:46some of the safeguards.
45:47And, again, I mentioned disease
45:49surveillance,
45:50but there will be a
45:51category code go into effect
45:52on January first of this
45:53year. Now I'm gonna kinda
45:55walk you through some of
45:55the challenges that with this
45:57approach. This happened a lot
45:58faster than any of us
45:59thought would happen in the
46:00field, that we would be
46:02paying for this technology,
46:03at this early stage. One
46:05is that how I scan
46:06a patient will markedly affect
46:08what I measure in that
46:09patient. And so if I
46:10this is data from Paradigm
46:12where patients got serial CT
46:13scans for clinical reasons separated
46:15by about three little more
46:16than three and a half
46:17years.
46:18And what you saw was
46:19vast differences
46:21in low attenuation plaques simply
46:22based on the tube potential
46:24of the CT. So for
46:25example, if you lower tube
46:26potential, the contrast or the
46:28density of contrast in the
46:29lumen goes up, and many
46:30of our pixels include both
46:32lumen and wall.
46:34And when you do that,
46:35what you measure in the
46:36lumen or sorry, in the
46:37in the wall of the
46:38artery also goes up called
46:39partial volume averaging. And because
46:41of that,
46:42we know that people who
46:43have high attenuation
46:45in the aorta
46:46have much less likely chance
46:48of having fibrofatty or low
46:50attenuation plaque measured. So details
46:52matter in how you scan
46:53patients. And you can imagine
46:54how this would matter if
46:55you got scanned
46:56twice with two different tube
46:58potential settings. And so we
46:59have to put some safeguards
47:01on this. And, this is
47:02the data I just mentioned.
47:03This is the seven study
47:04the seven scans. This is
47:05someone with obviously significant LED
47:08disease, but you can see
47:09the stenosis ranges
47:11across these software,
47:12significant differences in total plaque
47:14volume. And the important thing
47:16is how much plaque you
47:17measure differs significantly based on
47:19how much of the length
47:20of the coronary as you
47:21analyze.
47:22And each of these software
47:23has different extensiveness, which where
47:25they're analyzed. One will go
47:26to two millimeters. Another goes
47:28to one point five. Another
47:29goes to one point eight.
47:31And how they deal with
47:32artifacts is also very, very
47:35vendor dependent.
47:38And I say this because
47:39there are many studies now,
47:40and this is what I'm
47:41working on some of these
47:42to try to make sure
47:43that people who do this
47:44know what they're doing and
47:45they do this correctly. So
47:47we do know that medical
47:48therapy can change the progression
47:49of plaque. This is, again,
47:50an observational study we were
47:52involved with called Paradigm.
47:54These were patients who had,
47:55again, serial CT scans for
47:57clinical reasons, and we simply
47:59looked at patients who were
48:00on statins and who were
48:01not on statins. And what
48:03we saw was no over
48:04four years, again, using a
48:06plaque quantification
48:07software, are that patients who
48:09were on statins had a
48:10reduction in non calcified plaque,
48:12an increase in calcified plaque,
48:14a slight
48:16decrease in total plaque or
48:17percent atheromol volume, but that
48:18was pretty minimal.
48:20So we've we've learned that
48:22the plaque biology can be
48:24changed, and there are multiple
48:25clinical trials that are still
48:26doing this.
48:27But there's devil in the
48:29details that you have to
48:30be aware of, and so
48:31this this is what, you
48:32know, people,
48:33I think, have have have
48:34viewed as the potential
48:36benefit of plaque software. It's
48:38very challenging to look at
48:39scans side by side
48:41and visually know if someone
48:42has gotten better. Software can
48:44probably help us here,
48:46and this is just one
48:47example of someone who's gone
48:49from more non calcified plaque
48:51to blue here, more calcified
48:52plaque. And so this may
48:53be helpful, and this is
48:54why many drug trials
48:56are ongoing using CT as
48:58an endpoint,
49:00not for drug approval. It's
49:01not a valid surrogate for
49:03the FDA, but for
49:04biological
49:05insights as to whether the
49:06drug affects the biology of
49:08atherosclerosis.
49:09But to do this well,
49:10you have to be careful.
49:12And so what I mean
49:12by that,
49:13is there are challenges. I
49:15mentioned already the fact that
49:16you, have to scan these
49:18patients as exactly the same
49:20scanner,
49:21kernel reconstruction settings, slice thickness
49:24to potential
49:25contrast delivery,
49:28and that is something that
49:29is vitally important or you
49:30will get really wacky numbers.
49:33Now there some of these
49:34may be overcome by photon
49:35counting CT, and I think
49:36that's really the the future
49:38of using mono energetic
49:41imaging
49:42to standardize how we reconstruct
49:43images. But today,
49:46this is a real challenge
49:47to assessing disease progression. And
49:49I say that because some
49:50of the vendors will come
49:51to you and say,
49:52if you scan somebody and
49:54you do a plaque analysis,
49:56maybe you should repeat it
49:57in two to three years
49:58and see if they're doing
49:59okay.
50:00And some patients have been
50:01told that and will come
50:02to you and say, hey.
50:03Is my my heart disease
50:04getting better? You've got me
50:05on this combination lipid lowering
50:07therapy. You've got me on
50:07a GLP one and SGLP
50:09two. Is it getting better?
50:10It's a valid question.
50:12We do that in a
50:13lot of other disease states.
50:15But my point is it's
50:16not as easy as going
50:17downtown and just getting another
50:18CT and sending it for
50:19software analysis.
50:21And I tell you this
50:21because in clinical trials, we've
50:23done this test. We've gotten
50:25very different results
50:26from the same vendor simply
50:28based on the phase that
50:29they choose
50:30or based on the technologist
50:32at the vendor who ran
50:33it. I'll give you an
50:33example. We had a patient
50:34who had a serial CT.
50:36They ran it with two
50:37different
50:38technologists. One picked one phase,
50:39another picked another phase. One
50:41told us the patient progressed.
50:43One told us the patient
50:44regressed.
50:45Same patient.
50:47Same company.
50:48Same software.
50:49Imagine that across different software.
50:51So my point is if
50:52you're gonna use this for
50:53disease progression, the devil is
50:54in the details,
50:56and the software
50:57have not adjusted
50:59or counted for many of
51:00these
51:01these acquisition differences.
51:03Meaning, they're the outputs that
51:04you get are not
51:06different or modified or,
51:09you know, standardized based on
51:11two potential and other acquisition
51:13parameters.
51:14Now software I mentioned high
51:15risk plaque. Can we use
51:17software to help us, you
51:19know, better, you know, identify
51:20high risk plaque? I showed
51:21you the kind of simplistic
51:22way which we look at
51:23a plaque and say is
51:24it high risk. You know?
51:25And this is another area
51:26of investigation. This is a
51:28study just recently published called
51:29Emerald two, and this was
51:31a population of patients who
51:32had had a CT either
51:34one month or up to
51:35three years prior to an
51:37m MI, and they had
51:38invasive angiography
51:40to identify the culprit lesion.
51:41And they used AI. And
51:42they said, can you train
51:44an AI model to identify
51:45the culprit lesion?
51:47And what they found is
51:48they did a reasonable job,
51:50and what they found is
51:51that that Delta FFR CT,
51:52the change in estimated pressure
51:54across the lesion, higher plaque
51:56burden, again more total plaque
51:58volume and more low attenuation
52:00plaque volume.
52:01And this somewhat kinda clunky
52:03estimate of myocardial blood flow
52:04with CT, these tended to
52:06be the the five predictors
52:08of culprit lesions. And so
52:10this is very early work,
52:11but the software companies, I
52:13imagine, in the future will
52:14come to you and say
52:14we can not only assess
52:16high risk plaque,
52:17you know, that you see,
52:18but we can identify high
52:19risk plaque from an AI
52:21algorithm that may be more
52:22effective. We'll see.
52:23The last thing I wanna
52:24do before I close is
52:25just touch on some of
52:26the other features that are
52:27being investigated. These are kinda
52:28hot topics in the area
52:30of CT.
52:31And one is this question
52:32of can you also infer
52:33information about the degree of
52:35inflammation? I think all of
52:37us in the room are
52:37aware that inflammation drives,
52:40a large part of atherosclerosis
52:42and perhaps its progression.
52:44And investigators in Oxford have
52:45looked at the fat that
52:47immediately surrounds the proximal coronary
52:49arteries.
52:50And we know that in
52:52this scenario
52:53that you see, I'll just
52:55mention this briefly, but one
52:56of the things that you
52:56see on histology slides are
52:59that areas of patients who
53:00have significant atherosclerosis
53:03and or systemic inflammation,
53:05we see a reduction in
53:06the size of the adipocytes
53:08immediately near the coronaries.
53:10We see an increased water
53:12and inflammatory
53:13cell content
53:14and an increase on CT
53:16correspondingly of higher CT density.
53:20And so others have said
53:21why we already have this
53:22information in the CT scan,
53:23why don't we use it?
53:24Why don't we use this
53:25to assess? And so what
53:27we what is what has
53:28been shown here is that
53:29patients using a an AI
53:31tool,
53:32where they measure this density,
53:33this increased density around the
53:35coronaries, patients who have really
53:37high densities around their coronaries
53:39in this fourth quartile,
53:40and they've labeled this a
53:41PHY score, have a significantly
53:43increased risk of cardiovascular hard
53:45cardiovascular events. And this was
53:47in a very large population
53:48called Orphan published in Lancet.
53:50This is really the largest
53:51cardiac CT trial done to
53:53date.
53:54This is, you know, over
53:55thirty thousand patients. So
53:57more to follow. This is
53:58in its very early stages,
54:01and I do have some
54:02of the same concerns about
54:03this technology based on acquisition
54:05differences.
54:06The vendor says we account
54:08for all of those. Meaning
54:09if you scan someone at
54:10seventy kV,
54:11we scan someone at one
54:12hundred and twenty kV. This
54:13unlike our plaque colleagues, we
54:15account for that. But they
54:16don't really tell us how
54:17they do that. And so
54:18this is kind of an
54:18area of intensive investigation. Last
54:20thing I wanna say is
54:21that is the area the
54:22the field is changing.
54:23We're seeing this new technology
54:25which may overcome many of
54:26these limitations
54:28using photon counting CT. I
54:29won't we won't go through
54:30all of the physics of
54:31this, but suffice it to
54:32say it has significantly higher
54:33spatial resolution,
54:35which allows us to overcome
54:37some of these limitations.
54:38Pixel sizes are much smaller.
54:40Our ability to identify plaque
54:42and separate it from pericornary
54:44fat and calcium
54:46is improved. And what data
54:47has shown is that when
54:48sites that have photon counting
54:50CT, the referral rates to
54:51cath go down, the overall
54:53accuracy of patients who go
54:55to the cath lab go
54:55up mainly
54:57at an increased specificity.
54:59You know, CT does a
55:00pretty good job, just regular
55:01old CT,
55:03at ruling out high risk
55:04coronary disease. But our specificity
55:06goes up because we're able
55:07to see more of the
55:08lumen.
55:10And so this is something
55:10that may change some of
55:11these challenges that I just
55:13mentioned and this is just
55:14some of this data from
55:15this largest site. And in
55:16fact, it may change how
55:17good we are at identifying
55:18low risk plaque and this
55:19is a a paper just
55:20recently from radiology showing that
55:22using photon counting CT, you
55:24can actually see that the
55:25agreement
55:26for high risk plaque went
55:27up significantly
55:28as compared to standard, CT
55:30scans. And these were patients
55:31who got scanned on both
55:32scanner platforms.
55:34So the last thing I
55:35want to say is that,
55:35you know, there's are several
55:37clinical trials that are looking
55:38at screening CT. This is
55:39a very,
55:40controversial topic.
55:42You know, people who advocate
55:43for this will say, we
55:44already do this in so
55:45many disease states. We have
55:47transitioned to using imaging.
55:50We don't do a great
55:51job, by the way. You
55:52know you know, this is
55:53a recent publication, two thousand
55:55seventeen to two thousand twenty
55:56two, or four point six
55:57million patients in the US.
56:00One in five are on
56:01a statin prior to their
56:02event, and so this has
56:03been, you know, certainly,
56:05data that has armed people
56:06to say maybe we can
56:07use imaging to do better.
56:09It's not better risk scores.
56:11These risk scores do not
56:12perform well. This is our
56:13new PREVENT score across four
56:15different what you would think
56:16would be relatively like
56:18academic practices, and there was
56:20very, very poor discrimination,
56:22and poor calibration for who's
56:24actually having an event.
56:26And so can we use
56:26imaging? Wait and see.
56:29Calcium scoring you heard today
56:30does well. We know from
56:32large scale screening trials like
56:33SCAPIS in Sweden
56:35and the,
56:36Miami Heart Study that when
56:38you screen asymptomatic patients, you
56:39find a lot of atherosclerosis.
56:40About half of patients who
56:42were low risk had plaque,
56:44and this plaque is prognostically
56:46important.
56:47This is data from from
56:48Denmark showing that asymptomatic patients
56:50have more events if they
56:51have more of their atherosclerosis.
56:53So there are two large
56:54scale or actually four large
56:55scale trials. I'll mention two
56:57of them that are ongoing.
56:58This is SCOTHEART two. This
57:00is looking at six thousand
57:01patients in Scotland,
57:03and they're randomizing them to
57:04screening coronary CT
57:06versus usual care,
57:08is now fully enrolled.
57:10This is the transform trial.
57:11This is a trial that's
57:12randomizing patients who have metabolic
57:15syndrome or diabetes,
57:17to CT. They're doing serial
57:20CTs in this trial. They're
57:21using very aggressive prevention for
57:23PC. They're using LDL goals
57:25of less than thirty in
57:26those with high risk.
57:27And so these are these
57:28are trials that will be
57:29reported in the next three
57:30to four years,
57:32And we are a site
57:33for one of those trials,
57:34led by Pam Douglas, which
57:35is looking at young patients
57:37at low risk,
57:38who have atherosclerosis.
57:40This is not an outcomes
57:41trial, but it's looking at
57:42prevalence
57:43and, changes in risk factors.
57:45So I think I'll end
57:46and close to say, you
57:47know, in in twenty twenty
57:49five, CT has come a
57:50long way.
57:51It is at our institution
57:53has become really foundational to
57:54many, not just coronary imaging,
57:56but structural heart imaging.
57:58And I think it's more
58:00maybe more central to decision
58:01making in an era where
58:03we have nearly a dozen
58:04therapies that have been shown
58:05to reduce heart cardiovascular
58:07events. And so I think
58:08use the I use this
58:09plaque data with every patient
58:10that I see to assess
58:11pretest probability in chest pain
58:13patients, but also to refine
58:15the intensity
58:16of preventive therapy.
58:18I think it's exciting what
58:20we see from vendors to
58:21quantify this plaque that we
58:22currently manually would take me
58:24half a day to qualify
58:25to quantify. But I think
58:26while we're all enthusiastic,
58:29I think we're not excited
58:30about the price. And I
58:31can tell you from a
58:31cross sectional standpoint as a
58:32preventive cardiologist in my clinical
58:34world, I don't really need
58:36those plaque numbers to treat
58:37most patients.
58:38I I know how to
58:39treat them just looking at
58:40their CT scan. I probably
58:42will need it if you
58:43won't ask me if their
58:44disease is getting worse. So
58:46serial progression is probably the
58:47role for c for these
58:48plaque,
58:49software. They don't like to
58:50hear that because that means
58:51you're not gonna use them
58:52for the first scan.
58:54But I think that's kinda
58:56where we're moving, and I
58:56think the field has to
58:57get refined. The algorithms they're
58:59using are old.
59:00They're pre the transformer
59:02upgrades and in in deep
59:03learning.
59:05They the f data makes
59:06them lock them in. You
59:07can't update them on the
59:08fly. You have to resubmit.
59:10Hard flows never increase there.
59:12It never change their algorithm,
59:13and and and that's for
59:14that reason. So I think
59:15they need to get better,
59:16and I think they should
59:16probably should be retrained on
59:17full time counting CT.
59:19So thank you very much,
59:21and,
59:21and, again, thank you for
59:23this opportunity.
59:32Yes, sir.
59:43It was a very nice
59:45balanced presentation.
59:47Thank you. So the technical
59:48parts eventually may resolve, and
59:51you can address them. The
59:52biological part, you can't do
59:54anything with it. And plaque
59:55phenotype, you did we discussed
59:58that it's modifiable.
60:00It's also not a it's
01:00:01a dynamic process. That's right.
01:00:03Everything changes within a very
01:00:05short period of time, in
01:00:07fact, months. So
01:00:09is there any and then
01:00:09the pause based because of
01:00:11that, the positive predictive value
01:00:13of all of this is
01:00:13below two percent, which is
01:00:15our threshold for
01:00:16for doing things. So is
01:00:18this really going to fly
01:00:19anytime in the future even
01:00:21if the techniques get better?
01:00:22When you say It means,
01:00:25adjusting treatment based on plaque
01:00:27characterization.
01:00:29Well, I think, you know,
01:00:30I look at this as
01:00:30more as patient risk. And
01:00:32so I think when you,
01:00:33you know, the challenge I
01:00:34have in my clinical setting
01:00:36is, you know, we have,
01:00:37you know, who needs to
01:00:38go on sometimes lifelong expensive
01:00:39therapies. And so I think
01:00:41trying to match,
01:00:42you know, the intensity of
01:00:44treatment
01:00:45with the risk of the
01:00:46patient, I think, is what
01:00:47we're all kinda aiming to
01:00:48do. And we don't wanna
01:00:49overtreat patients,
01:00:51but I think right now
01:00:52when you look at this
01:00:53this this summation of, of
01:00:55outputs that you have, such
01:00:56as calcium scoring, stenosis,
01:00:59plaque extensiveness
01:01:00or burden, to include high
01:01:02risk plaque being a part
01:01:03of that, that I think
01:01:05you can actually see that,
01:01:06you know, there is pretty
01:01:08wide differences in patient individual
01:01:10patient risk. Now I think
01:01:11where people get hung up
01:01:12is that you're right. I
01:01:13mentioned in, you know, if
01:01:14you look at just high
01:01:16risk plaques individually,
01:01:18most of them do not
01:01:19cause events.
01:01:20They do lack specificity,
01:01:22but we know as a
01:01:23patient level, these are also
01:01:24patients more likely to develop
01:01:25future high risk plaques,
01:01:27and that biologically they are
01:01:29different. And so I think
01:01:31it is helpful potentially
01:01:33to to to use these
01:01:34therapies. We don't have a
01:01:34lot of great data on
01:01:35how these therapies change plaque
01:01:37biology. We know from PCSK
01:01:38nine data that you see
01:01:40regression of noncalcified plaque.
01:01:42You see improvement and reduction
01:01:44in high risk plaque, and
01:01:45we know they also reduce
01:01:46events. And so I think
01:01:48that's the rationale for doing
01:01:49that, but we don't have
01:01:50data.
01:01:51Now well, am I you
01:01:51am I waiting for that
01:01:52data? No. And if I
01:01:53have a patient at high
01:01:54cardiometabolic risk who has a
01:01:55lot of high risk plaque,
01:01:56I'm probably gonna use an
01:01:57SGLT two inhibitor,
01:01:59GLP one
01:02:00until that data helps helps
01:02:02change my mind. Yep. Yes,
01:02:04sir. Thank you, Brent.
01:02:07Thank you for a nice
01:02:08thought. Yeah. And, also, for
01:02:10and, one of the most
01:02:12honest self reflections on their
01:02:14own field in the era
01:02:16of AI that I've seen.
01:02:17Really nice. Thank you.
01:02:19Charlie Taylor was here a
01:02:20few weeks ago from Artflow
01:02:22giving us a talk. A
01:02:23lot of what you're saying
01:02:24is true about the algorithms
01:02:25of things. One of the
01:02:26things that I I don't
01:02:28know if all these imaging
01:02:29parameters are always reported
01:02:31at use in the different
01:02:32softwares. Are they, like Mhmm.
01:02:34All this Well, I mean,
01:02:35the the the metadata within
01:02:37the DICOM,
01:02:38data is there.
01:02:40And that that is set
01:02:41up also proactively by this
01:02:43by the vendors,
01:02:44and so they they they
01:02:46have that data. The challenge
01:02:47is that if you look
01:02:48at
01:02:49I'll use the fat attenuation
01:02:50folks, and and their whole
01:02:52software was validated on a
01:02:54hundred and a hundred and
01:02:55twenty kV. That's it. They
01:02:57will run their software on
01:02:58all kVs.
01:02:59And so the algorithm does
01:03:01not
01:03:02it's not been trained using
01:03:04those kVs,
01:03:05and it's not really tuned
01:03:07and
01:03:08and and really, you know,
01:03:09adjusted for those differences. And
01:03:11that's kind of the same
01:03:11for plaque. Right? They if
01:03:12you look at most of
01:03:13the plaque vendors, they give
01:03:14you outputs based purely on
01:03:16houndsealed units,
01:03:18and they don't adjust their
01:03:19thresholds based on the k
01:03:21v, even though we know
01:03:22they should.
01:03:23And I think in the
01:03:24future, you know, the solution,
01:03:25if you talk to, you
01:03:26know, many of our, you
01:03:27know, like, Cynthia McCullough's of
01:03:29the world is it is
01:03:30will be monoenergetic
01:03:31reconstructions
01:03:32that can standardize this. Very
01:03:34little differences in software when
01:03:36you run them on the
01:03:36same monoenergetic
01:03:37images. That's what's gonna be
01:03:38a help hopefully, to help.
01:03:39Well, that's totally strange. You
01:03:40can see where you can
01:03:41fix that now. But what
01:03:43it could do, and I'm
01:03:44curious if you've seen this
01:03:45or even if you add
01:03:46it, which you find it
01:03:48helpful,
01:03:49is it might algorithm might
01:03:51be able to characterize
01:03:52uncertainty.
01:03:53That's right. Give you bounds
01:03:55on things. And so
01:03:56here's maybe the flack,
01:03:58but it's within this confidence
01:04:00bound. Would you find that
01:04:01helpful? Or
01:04:03We we've had this discussion
01:04:04is this, you know, level
01:04:05of diagnostic certainty.
01:04:07I actually kind of am
01:04:08a fan of it. I
01:04:09don't know. We've got some
01:04:10others in the room. I
01:04:10mean, when I say even
01:04:12for things like stenosis. Right?
01:04:13I mean, I say there's
01:04:14a severe stenosis. Should I
01:04:15qualify that with a level
01:04:16of diagnostic certainty?
01:04:18I think
01:04:19many of us, like me,
01:04:20would like that to put
01:04:22that in my reports so
01:04:23that they I I try
01:04:24to convey that with words,
01:04:25but I think that would
01:04:26be helpful. I think some
01:04:27of the end users might
01:04:28not like that. And so
01:04:29my interventional cardiologist
01:04:31would probably and they have
01:04:32they have told us that.
01:04:33So I it's not something
01:04:34we've started using. A AI,
01:04:36I think people want black
01:04:37and white answers usually, but
01:04:38I don't know. I but
01:04:39I think you're right. Now
01:04:40the challenge is artifacts too.
01:04:41I have seen we have
01:04:42run analysis for studies on
01:04:44four different vendors,
01:04:46and their rejection rates are
01:04:47all variable. They don't reject
01:04:49the same cases, and they'll
01:04:50sometimes analyze cases. And you're
01:04:52like, I have no idea
01:04:53how they analyze that vessel.
01:04:55And they have they say,
01:04:56well, they have a motion
01:04:57correction algorithm. They can find
01:04:58the lumen. I can't even
01:04:59see the coronary.
01:05:01So that's a challenge as
01:05:02artifacts is another challenge. There's
01:05:04no consistent approach to that.
01:05:11Phenomenal talk as usual. Thank
01:05:13you very much. I'm gonna
01:05:14put you on the spot
01:05:15because you were talking about
01:05:18the validity of calcium scoring
01:05:21and that people who don't
01:05:23have,
01:05:24who have a calcium score
01:05:25of zero
01:05:26likely don't have any
01:05:29coronary disease that is
01:05:31actionable or severe. Well, I
01:05:33think what what I what
01:05:34I said is they have
01:05:35we know from large populations
01:05:36they have low event rates.
01:05:37I showed you a case
01:05:38where that was not the
01:05:39case. That that was exactly
01:05:40my you know, that's where
01:05:41I'm you know, where I'm
01:05:42heading. Yeah. The
01:05:44calcium scores have become extremely
01:05:45popular, and see we see
01:05:47them being ordered by our,
01:05:49you know, primary care providers
01:05:50on all sorts of people,
01:05:51including thirty year olds.
01:05:55Yeah. And my question for
01:05:57you
01:05:58is how much,
01:06:01well, how much
01:06:03does a calcium score of
01:06:04zero in a high risk
01:06:06patient
01:06:07who is thirty years old
01:06:09count?
01:06:10Is it valuable, or should
01:06:12we just assume this patient
01:06:14may have more than just
01:06:15calcified
01:06:16plaque if they are a
01:06:17high risk patient
01:06:18and proceed and do a
01:06:20coronary CTA if we're looking
01:06:21for
01:06:23plaque.
01:06:24Yep. So we know I
01:06:25just I just put up
01:06:25a slide here. I kinda
01:06:26went through this quickly for
01:06:27time, but in symptomatic patients,
01:06:29it's unquestionable.
01:06:30CTA is better.
01:06:33It's I I get in
01:06:34these arguments with doctor Budoff
01:06:35and with Karim Nasir, and
01:06:37they're lovely people.
01:06:38And I my career started
01:06:40in calcium scoring.
01:06:42But when I say that,
01:06:43they get mad at me.
01:06:44I mean, they literally really
01:06:45get mad at me.
01:06:46But it's unquestionably better, and
01:06:48this is data from from
01:06:49PROMIS showing the hazard ratios
01:06:50for an abnormal calcium score
01:06:51versus abnormal CT. In asymptomatic
01:06:53patients, you will find,
01:06:56and this is from Scapis,
01:06:57in patients with calcium scores
01:06:58of zero, they had about
01:07:00six percent of people that
01:07:01had plaque.
01:07:02And the more risk you
01:07:04have individually, like family history
01:07:06and risk factors, that number
01:07:07goes up.
01:07:08And so you absolutely are
01:07:10missing disease. So CT is
01:07:11unquestionably going to be better,
01:07:13and we know that from
01:07:14this group just published a
01:07:15paper in JAMA this week.
01:07:17I didn't make a slide
01:07:18for it. And they showed
01:07:19that. They showed that the
01:07:20plaque findings, not quantification of
01:07:22plaque.
01:07:23The problem is people mix
01:07:24this up. They think if
01:07:25I'm talking about screening, CT,
01:07:26I must be
01:07:27wanting to do quantification. No.
01:07:28Just stenosis, high risk plaque,
01:07:30CADRADS two was better than
01:07:32calcium scoring. The problem is
01:07:33it was just a little
01:07:34bit better.
01:07:36There's a paper coming out
01:07:37next week in JAC Imaging
01:07:38that looks
01:07:39at confirmed two registry that
01:07:41also shows that plaque quantification
01:07:44was barely better than what
01:07:46we're already doing in an
01:07:47asymptomatic or in symptomatic population.
01:07:49So the question is made
01:07:50is not is it better,
01:07:51it's by is it worth
01:07:52it? And that's the real
01:07:54question is how much better?
01:07:55And so, you know, I
01:07:57think CTA is unquestionably better
01:07:58and and and people it
01:07:59may be better in those
01:08:00younger higher risk patients where
01:08:01you're gonna find this type
01:08:02of plaque. The fear is
01:08:04that I had this come
01:08:04up actually this week in
01:08:05my own clinic is that
01:08:06someone's gonna, at forty, get
01:08:07a normal calcium score. They've
01:08:08got an LDL of a
01:08:09hundred and eighty and say,
01:08:11I'm good.
01:08:12And that's what happened with
01:08:13the patient I just showed
01:08:14you. And that's the fear
01:08:15in those patients is that
01:08:17you shouldn't,
01:08:18you know, I think, overly
01:08:19reassure those people. Yeah.
01:08:21Very cool.
01:08:23This was an excellent talk.
01:08:25I really enjoyed it. I
01:08:27in a practical sense,
01:08:29how much these indices, my
01:08:31question, is gonna,
01:08:32be helpful.
01:08:34Not infrequently, we see people
01:08:36having actually high risk plaque
01:08:37such as napkin sign, but
01:08:38they have a normal FFR.
01:08:40Now if I have a
01:08:41patient with abnormal FFR, I
01:08:43know that if he has
01:08:44a chest pain, a stenting
01:08:46gonna help him.
01:08:47But if I have a
01:08:49high risk,
01:08:50plaque Mhmm. Do you have
01:08:52evidence that actually stenting that
01:08:54plaque is gonna help
01:08:56help this individual?
01:08:57I mean, that that's the
01:08:58question of practical Yeah. So
01:09:00you're asking a million dollar
01:09:01question there.
01:09:02You know, and and to
01:09:03be honest with you, we
01:09:04get that patient question from
01:09:05patients and interventionists.
01:09:07If you recall, there was
01:09:08a similar clinical trial published
01:09:10this past year and a
01:09:11half called the PREVENTS study
01:09:14that defined high risk plaque
01:09:15invasively
01:09:17by OCT.
01:09:18And if they were high
01:09:19risk, they stented them, and
01:09:20there was a benefit in
01:09:21that trial.
01:09:22We have no data that
01:09:23that that that that was
01:09:24a small trial. It was
01:09:25more of a proof of
01:09:26concept trial. So I think
01:09:28we have no data that
01:09:29preemptive stenting,
01:09:31is beneficial.
01:09:33I think, you know, I
01:09:34I would favor much strongly
01:09:35favor medical therapy. So I
01:09:37would not use high risk
01:09:38plaque to tell someone they
01:09:40need an intervention.
01:09:42People, you know, there again,
01:09:43there are people out there
01:09:44who think that as our
01:09:45stents become more like balloons
01:09:47and become
01:09:48safer potentially,
01:09:50maybe there
01:09:51are proxylid lesions that are
01:09:53just so high risk we
01:09:54can't not treat them, but
01:09:55that's that's that got has
01:09:56got to be proven with
01:09:57clinical trial data. So that
01:09:58is not today. I would
01:09:59say use it to refine
01:10:00or intensify therapy med medically.
01:10:03But you know the high
01:10:04risk plaque. High risk
01:10:05plaque, but you have a
01:10:06plaque.
01:10:07You will less intensely treat
01:10:09that patient? No. I would
01:10:10I would I would more
01:10:11I I tend to, you
01:10:12know, again, conventionally address risk
01:10:14with intensity of treatment. So
01:10:16if I have someone who
01:10:17has high risk atherosclerosis,
01:10:19a lot of atherosclerosis or
01:10:20both, I would treat them
01:10:21more intensely, but I would
01:10:22not use that in decision
01:10:24making for revascularization
01:10:26for particularly for chronic coronary
01:10:28disease where there's really not
01:10:29a lot of survival benefit.
01:10:30Right? You would only do
01:10:31that for symptom benefit mostly.
01:10:33Now maybe that'll change in
01:10:34the future,
01:10:36you know the Scott Arp
01:10:37two people tell me they're
01:10:38finding lots of scary looking
01:10:39disease and these people that
01:10:40are biking and hiking and
01:10:42feel great.
01:10:43We'll see what happens to
01:10:44them and what they do
01:10:45with that information.
01:10:48You know, there's also a
01:10:49fear that, you know, we
01:10:50do a lot of calcium
01:10:51scoring, but CT, if you
01:10:52do that in a screening
01:10:53population, you're gonna do a
01:10:54lot more casts and things
01:10:55like that. And some people
01:10:57said the opposite would happen
01:10:58because we see these high
01:10:59calcium scores, most of them
01:11:00will not have stenosis That's
01:11:02severe. Maybe it would reassure
01:11:03people. I don't know the
01:11:04answer to that.
01:11:06So before we close the
01:11:07program, I wanted to invite,
01:11:10Barry's widow, Renee, to come
01:11:11up and say a few
01:11:12words.
01:11:24Thank you.
01:11:29That was brilliant. Oh, thank
01:11:30you so much.
01:11:38So it's very difficult following
01:11:40the brilliance of doctor Valine.
01:11:44And so my comments
01:11:45will be on a very
01:11:47different scale.
01:11:51I want you to know
01:11:52that I feel honored to
01:11:53be standing here before you,
01:11:56Barry's colleagues
01:11:58and friends.
01:12:00And it is with gratitude
01:12:01from our family,
01:12:04to doctor Eric Velasquez
01:12:07and to Al,
01:12:09Stanusis
01:12:10and to the other members
01:12:12of the Department of Medicine
01:12:15for the creation
01:12:17and naming
01:12:18of the symposium,
01:12:20the Barry
01:12:21Zarratt
01:12:22Symposium
01:12:23and Grand Rapids.
01:12:26Barry would have been deeply
01:12:29humbled
01:12:30by the tribute
01:12:32from his colleagues and friends
01:12:34whom he deeply respected,
01:12:37cherished,
01:12:38whose presence he admired,
01:12:41and from whom he received
01:12:44inspiration.
01:12:47As you may know,
01:12:49Barry was a man of
01:12:51many talents,
01:12:54and he published,
01:12:55in addition
01:12:57to the more than three
01:12:58hundred articles
01:13:00and
01:13:01the
01:13:02four books
01:13:03on cardiology.
01:13:05He also published
01:13:07three books of poetry.
01:13:10And in keeping
01:13:11with the subject of imaging,
01:13:15I want to read
01:13:16a poem that Barry wrote.
01:13:19It is called
01:13:20Nuclear Images.
01:13:23It comes from the third
01:13:25book
01:13:26called A House of Many
01:13:28Rooms,
01:13:30and it, was published in
01:13:32two thousand twenty one.
01:13:37Nuclear images.
01:13:40My life,
01:13:41a montage
01:13:43of images
01:13:45in the
01:13:46lab, nuclear images,
01:13:49heart scans showing blood supply
01:13:52after isotope injected.
01:13:56First done when young,
01:13:58career beginning,
01:14:00initial success.
01:14:04Now
01:14:05daily practice.
01:14:07Many patients studied, all awaiting
01:14:10answers,
01:14:12some anxious,
01:14:15some resigned,
01:14:18some seeming not to care.
01:14:21Images read,
01:14:23reports written,
01:14:25work done.
01:14:27All the while,
01:14:29knowing little of those imaged,
01:14:33their families,
01:14:36loves,
01:14:38woes,
01:14:39occupations,
01:14:41their life's poetry.
01:14:45Will the tests
01:14:46trigger
01:14:47new treatment,
01:14:49concern,
01:14:51alarm,
01:14:54fear,
01:14:55or
01:14:57relief.
01:15:00Days follow days.
01:15:02Pixels
01:15:03follow pixels.
01:15:05Images
01:15:07follow images.
01:15:08Schedules
01:15:09filled and refilled.
01:15:13Patients return to change lives.
01:15:16Unaltered
01:15:17readers
01:15:19remain in offices,
01:15:21viewing more images
01:15:23in two dimensions, unaware
01:15:27of the humanity
01:15:29behind each study,
01:15:31quenching thirst from half filled
01:15:33glasses.
01:15:36Nuclear images,
01:15:38my profession.
01:15:40Nuclear
01:15:41patience,
01:15:42my soul.
01:15:45Thank you.
01:15:51Alright. Thank you.
01:15:53And thank you, doctor Valence,
01:15:55for a wonderful presentation.
01:16:00Alright. So lunch is served,
01:16:01and please take time to
01:16:03see the posters.