0:02

Hello and welcome to Noon Conference hosted by M R I Online Noon Conference

0:06

connects the global radiology community through free live educational webinars

0:10

that are accessible for all and is an opportunity to learn alongside top

0:14

radiologists from around the world.

0:16

We encourage you to ask questions and share ideas to help the community learn

0:20

and grow.

0:21

You can access the recording of today's conference in previous noon conferences

0:25

by creating a free m r I online account.

0:28

You can also sign up for a free trial of our premium membership to get access to

0:32

hundreds of case-based microlearning courses across all key radiology

0:36

subspecialties. Today we are honored to welcome Dr.

0:40

Moritz Hoffman for a lecture on assessment of fractional flow reserve from

0:44

coronary CT an geography. Dr.

0:47

Hoffman is a resident and clinician scientist in radiology at the University

0:51

Medical Center in Mines Germany.

0:53

His special focus lies on cardiovascular imaging and the integration of

0:58

artificial intelligence into radiology workflows. Dr.

1:01

Hoffman's passions lies in bridging the gap between cutting edge technology and

1:06

clinical practice.

1:07

He has been actively involved in validating and integrating AI algorithms into

1:11

cardiovascular workflows with the aim of enhancing diagnostic accuracy,

1:15

efficiency, and patient outcomes. At the end of the lecture, please join Dr.

1:20

Hoffman in a live q and a session where he will address questions you may have

1:23

on today's topic.

1:25

Please remember to use the q and a feature to submit your questions so we can

1:28

get to as many as we can before our time is up. With that,

1:31

we are ready to begin today's lecture. Dr. Hoffman, please take it from here.

1:36

Well, thank you very much for that kind introduction.

1:39

I'm happy to be here as well. And, um,

1:41

I will focus on fractional flow reserve from coronary CT angiography today.

1:48

So let's look at the agenda, what we are gonna tackle today. Um,

1:52

first up we're gonna start with, uh, the rationale for C T F F R.

1:57

To do that, um,

1:58

we're gonna review current C C T A evidence and look at challenges that we face

2:02

in C C T A and potential solutions and how C T F F R is one of the

2:07

potential solutions.

2:09

And then we will sort of narrow in on F F R and review technical aspects

2:14

of f r and also clear up all these acronyms like C F D

2:19

F F R, ml, F F R. All these will then, um,

2:23

mean something to you after this talk. And of course,

2:26

while it's a still a good thing, it does have some CHA challenges as well.

2:31

And we'll look at those and at potential solutions. And then in the end,

2:35

we'll look at the clinical impact of C T F F R and uh,

2:38

and clear up when to use F F R and how to use F F R in the clinical reality.

2:44

So let's move right in with current C C T A evidence. And throughout this, uh,

2:48

talk, you'll see, uh, landmark AR articles here on the left side,

2:53

and they'll have a small QR code here on the bottom that you can scan with your

2:57

camera app on your cell phone.

2:59

And that will take you directly to the full article. And, um,

3:03

I believe every, uh,

3:07

publication except for one will be, uh, open access.

3:10

So that should get you quite far. So this is, um,

3:14

the confirmed, um, registry study that was published, uh, back in 2011,

3:19

so quite a while back. And, um,

3:22

really what this showed for C C T A is that C C T A has incremental value

3:28

over, um, L V E F alone that you can do by echo. And,

3:33

um, if you look at it, it's not only that detect coronary artery disease,

3:37

but you also detect high risk or no high risk obstructive cardio,

3:43

coronary artery disease.

3:45

So what we can say is plaque plus stenosis is the prognosis,

3:49

and that is, so that is a key finding that will be reiterated later on in,

3:54

in, uh, more advanced studies in the,

3:57

that have been published in the last few years.

4:02

One of those is, uh, the Promise study, which was published in 2015,

4:07

and that was a randomization between anatomical and functional testing.

4:11

Functional testing in this scenario is stress, E C G,

4:15

nuclear stress testing or stress echo and anatomical testing

4:20

is coronary, uh, CT angiography, so it's C C T A.

4:24

And what we can see here is clinical outcomes after a median follow-up of two

4:29

years. Were similar for both strategies,

4:32

so we can conclude this is a safe method versus functional testing.

4:36

This is basically the non-inferiority stage.

4:41

Moving further ahead, um, the well-known Scott Hart trial, um,

4:46

well, and the five-year outcome of it, um, was reported, um,

4:51

also in, in the New England Journal of Medicine. And this, uh,

4:55

showed a lower risk of death from coronary heart disease and non-fatal

4:59

myocardial infarction when we use C C T A compared to

5:04

the standard care alone. And standard care in,

5:08

in this scenario was exercise E C G and symptoms and other clinical evaluation.

5:13

Um, if you look at this closely,

5:16

you sort of have to look at how this trial was designed because they had, um,

5:21

they chose a path and then C C T A was added or not added, basically.

5:26

Um,

5:27

so the benefit that you see here is basically is really

5:32

just the added C C T A and this reiterates the, uh,

5:37

confirm registry results because you can see C C T A what C C T

5:41

A brings to the table here is, um,

5:45

it is the anatomical information on the plaque and that leads to

5:51

initiation, um, of, of, uh, statins, for example,

5:55

which then, uh,

5:57

improves prognosis at leads to lower event rates.

6:05

All right. And then this is, um, the Ischemia trial.

6:08

And if you know the Ischemia trial, then you probably, you know,

6:12

that it doesn't have that much to do with C C T A, but what this was, was a,

6:17

uh, randomization between conservative and invasive strategy.

6:22

And key detail here is that it was the initial, uh,

6:25

therapy that was randomized. So, um, the,

6:30

the conservative group was OPT optimal medical therapy,

6:34

but it could also be C C T A in there as well.

6:38

And what we see here is you see similar incident rates,

6:41

again for conservative and invasive strategy,

6:44

and you can use C C T A as a gatekeeper for invasive strategy.

6:49

So conservative strategy plus C C T A can, um,

6:54

can work and can be used as a gatekeeper to prevent unnecessary

6:59

icas. And that is one key finding here as well.

7:03

And then basically at the end of that line is the discharge trial,

7:07

which was published, uh, just last year. And that was a,

7:10

a fairly straightforward randomization between C C T A and invasive

7:16

procedures, invasive coronary angiography on the other side.

7:19

So in here we see a, um,

7:23

we see 48 months of 40 years of follow up and similar may rates for both

7:27

strategies. Um,

7:28

but we see fewer procedure related complications in the C C T A group.

7:34

So now we have basically proven at this point, um,

7:37

that C C T A is feasible, it is non-inferior,

7:41

and it is safe to defer based on C C T A, um, and we can,

7:46

uh, lower the rate of procedure related, um, problems.

7:53

And that of course prompted them guideline recommendations from major

7:57

guideline, from major associations like the NICE guidelines,

8:01

the E S C guidelines and the A H a. And, um, at this point,

8:05

you could just close this talk and say, okay, that's all we need. This is great,

8:08

but of course we all know that's not really the case. Um,

8:13

some problems remain

8:16

and there's really two big parts that remain that are problematic with C C T

8:21

A one, uh, big part is the image quality.

8:24

And if you look at the top pictures here,

8:26

you can see the three D visualization of the right coronary artery here.

8:31

And you can see a, uh, curved curin projection of that same artery here.

8:36

And there's a segment that is just non invaluable because of a step artifact

8:40

that is due to image quality and or to motion. Um,

8:45

and you can can overcome this except for

8:49

faster imaging or whatever, but this will always be a sort of a problem.

8:54

On the bottom here you can see a heavily calcified left anterior descending

8:58

artery, um, with a big plaque here. And you can appreciate the calcium blooming,

9:03

which basically narrows the lumen beyond, um,

9:08

beyond really accessibility. And if you look on the right,

9:12

it's the same patient, same, same scan,

9:15

but it's scanned with a higher resolution, uh, on a,

9:18

on like the latest generation of scanners.

9:21

And you can see the higher resolution really helps you reduce calcium blooming,

9:25

but it also introduces new problems like noise.

9:28

And we will look into that later on as well.

9:30

But really what we wanna focus on here is not much the image quality

9:35

part of, uh, part of challenges for C C A,

9:39

but the diagnostic uncertainty. And really, if you,

9:42

if you only read one of those studies that I'm showing here,

9:45

then please read this one because this is really,

9:47

captures really the essence of why we need

9:51

F F R and why we need C T F F R as well.

9:55

Because this study compared for invasive coronary angiography and for

10:00

C C T A,

10:02

it compared patients without coronary artery disease and with relevant coronary

10:07

artery disease. And then evaluated does,

10:11

is there a significant stenosis based on flow?

10:15

So really we are comparing anatomical imaging with functional

10:20

significance of, of stenosis.

10:23

And this is the first time you will see F F R here 'cause that is the gold

10:27

standard.

10:28

And what we can get from these points that are data points that are basically

10:33

spread out over the entire graph is that we are not able to

10:38

determine functional significance based on anatomical imaging.

10:42

And that is regardless of method, that is Q C A.

10:45

So quantitative invested coronary angiography or C C T A.

10:52

So there's a problem. And how are we gonna solve that? Well,

10:58

whenever there's a problem for C C T A, you can always look at, okay, what are,

11:02

what are the invasive coronary angiography people doing?

11:05

And they have been doing something for quite a while.

11:07

Now this is the first publication that validated this technique,

11:11

which is fractional flow reserve. So really how did you measure it?

11:17

You basically take a G wire and that g wire has a

11:21

distal and a proximal pressure sensor,

11:24

and you push that G wire over the stenosis and then you can measure

11:29

distal coronary pressure and proximal coronary pressure.

11:33

And then you can see if that, if that stenosis causes a pressure drop.

11:39

Now this is very important. You have to do this during maximum hyperemia.

11:43

So usually what what you do when you do it in,

11:46

in the cath lab is you inject TracRac, coronary denin, um,

11:51

in order to induce maximum hyperemia that is necessary because otherwise

11:56

you cannot, uh,

11:58

you cannot estimate flow from that measurement 'cause you can only measure

12:01

pressure,

12:02

but hyperemia equalize the coronary resistance and thereby you can measure

12:07

flow because it becomes proportional. And the same sense,

12:11

you can also account then for collateral flow and for myocardial demand.

12:17

So this is quite a, quite an interesting, uh, technique. And um,

12:23

this was of course then validated in extensively. And I will show you the,

12:27

the two landmark trials, the fame one and the FAME two trial. Um,

12:32

this was back in 2009, so quite a while back. Um,

12:36

and they randomized F F R guided P C I versus angiography guided P

12:41

C i. So comparing functional measurement to anatomical measurement basically,

12:46

but still in a catheter setting.

12:49

And what we can see here is that there was a 28% risk reduction in

12:54

maize when we used F F R guided P C I and even higher risk

12:59

reduction of death or heart attack. And F F I guided P C I.

13:03

So what we can take from that is basically the,

13:08

the takeaway is we do not intervene for stenosis that have a normal

13:13

F F R and that this might sound counterintuitive if you

13:18

look at this graph right here,

13:20

but that's because the angiography guided P c I

13:25

had a higher rate of revascularization of non,

13:29

or we should say at non necessary revascularization,

13:33

which comes then with an increased risk of thrombosis or osis,

13:38

which then leads to another event or even MACE in

13:43

that case. So really what we can take away from,

13:46

from this study is if we use F F R and it

13:51

has a good F F R A good fractional flow flow reserve,

13:56

then we have, we are pretty confident that we do not have to intervene.

14:02

And FE two basically looked on the, on the, uh,

14:05

from the problem from the other side. Um,

14:08

and did F F R guided P C I plus optimal medical therapy or

14:14

optimal medical therapy alone.

14:16

Because really if you followed the FAME one trial,

14:20

you could just drastically put it, okay, then just don't do P C I at all.

14:25

Um, because that's,

14:27

that's basically putting it very drastically.

14:30

But that's really not the point because if you could look at this here, um,

14:34

medical therapy will have a higher incidence rate than P C I guided by F F

14:39

R. So while FAM one said, do not intervene,

14:44

stinosis higher than point A F F R,

14:47

frame two confirmed that we do have to intervene for stenosis that are have an F

14:52

F R of lower than 0.8.

14:55

And you can see their risk reduction is quite dramatic.

14:58

And hospital REM administration with urgent revascularization.

15:03

All right, so how can we use that in ct?

15:08

Because this is also, this is still with, um,

15:11

putting in uh,

15:15

intra coronary adenosine and inducing hyperemia and that's all

15:20

not possible really in ct, right? 'cause we are doing a static measurement. Um,

15:25

but what we can do is basically simulate invasive F F R.

15:30

And this is something that, that has been done in the aerospace industry,

15:35

for example, um, quite extensively and has been done for a long time as well.

15:39

Like you can see here on that fighter jet, before that jet is built,

15:44

you look at how the airflow is gonna be,

15:46

how the aerodynamics are gonna be and if that thing can even fly or not.

15:51

So basically it's taking that idea and applying it to

15:56

a CT scan of the coronary arteries.

16:02

And that is you take the CT data and then you build a three D model

16:07

comparable to that of that fighter jet.

16:10

And that is your anatomic model.

16:13

And then you take physiological data and that is basically this

16:18

complicated equation.

16:19

Here is a NEA stokes equation that treats blood as a

16:23

non-compressible fluid and then a supercomputer is used

16:28

in order to generate this C T F F R.

16:32

So key term here is computational fluid dynamics.

16:36

That's the C F D abbreviation that we saw all the way in the beginning.

16:42

So this is, this allows us to estimate f r

16:46

based on a simple static, um,

16:50

CT scan of the coronary arteries.

16:53

Of course this had to be validated and uh,

16:55

one of the first studies to validate that was the discover flow study and

17:00

that randomized C T F F R versus standard of care C C T A

17:05

and all of these studies that validated C T F F R will obviously have

17:10

invasive F F R as the gold standard to compare to.

17:14

So if you look at this at the red dash line, that's C C T A alone.

17:18

So the normal, um, quantification of stenosis, um,

17:22

per vessel and per patient. And the solid blue line will be F F R ct.

17:27

And you can see the marked increase in sensitivity and specificity, um,

17:33

for both per vessel and per patient.

17:37

So while this was, this was only, uh, quite a few patients, uh,

17:41

so it was a hundred, about a hundred patients. Um,

17:45

this was confirmed in similar de similarly designed, but a lot bigger, uh,

17:50

patient groups as well. Uh,

17:53

a couple years later we can see this basically the same graph per patient per

17:57

vessel.

17:58

And you can again see that C T F F R outperforms the standard of care.

18:04

And this is the next trial,

18:08

which is basically the biggest of these validation trials. Um, and you can,

18:13

again,

18:14

it showed marked increase in specificity versus C C T A and it showed

18:19

a comparative invasive F F R. It only showed a small bias, um,

18:24

towards lower values.

18:25

So mean difference between invasive F F R and C T F F R

18:30

uh, was 0.03, which is quite good. So we can,

18:36

we can be quite confident that the cutoff values can be similar, right?

18:43

So to move further ahead, um, the platform, uh, trial was designed,

18:48

that was also a randomized controlled trial. Um,

18:51

and this showed no adverse events in patients who had no ICA two

18:56

CT F F R due to CT F F R results. And now this, this,

19:01

we need to take a little bit apart 'cause this study design is somewhat

19:05

complex. It had two cohorts cohort, um,

19:08

one was the planned invasive cohort and one was planned non-invasive testing

19:14

based on the choice of local clinicians. Uh, prior to enrollment of this study,

19:18

both cohorts were then split into standard of care,

19:23

basically what the clinicians had planned or C T F F R

19:28

guided care. And then, then there was a 90 day follow up.

19:32

So even if, even if clinicians had planned invasive testing,

19:36

they would've undergone C T F F R. And if C T F F R was normal,

19:41

then there was,

19:42

it was deemed no need for I C A because of no obstructive coronary artery

19:47

disease. And then the patients were deferred basically. And after 90 days,

19:51

that was safe. So again,

19:54

this is very important evidence that showed C T F F R can safely defer, uh,

19:59

from, uh,

20:00

invasive correlation and the secondary analysis,

20:05

um, even, uh, more evidence for, for,

20:10

uh,

20:11

for C C T for C T F F R was found seeing that we have

20:16

lower cost, and again,

20:17

same study design 'cause it's the secondary analysis and the planned invasive

20:22

group, C T F F R was able to defer, um,

20:26

invasive procedures and therefore cost obviously dropped.

20:31

And then what they also reported was an increase in

20:36

quality of life after, uh, in,

20:39

in the C T F F R group. And really you can't be,

20:44

we can't be sure why that is because there was a non-randomized trial and was

20:49

unblinded. Um, but um,

20:54

it might be due to more frequent use of revascularization that were detected

21:00

in the F F R C T group because this was the planned non-invasive group.

21:04

And if they went with usual care,

21:07

they most likely got optical medical therapy and there might have been

21:11

a missed significant stenosis that was,

21:15

would've been discovered with CT F F R. All right.

21:20

Then the last of these of these, uh, validation trials,

21:25

um, or probably not the last,

21:28

but one of the latest is, uh, the Pacific trial,

21:32

which was, uh,

21:34

comparison between C T F F R and functional imaging and invasive

21:40

F F R. And right here we have everything put together.

21:44

Basically every functional imaging that you can imagine, um,

21:49

it's PET ct, it's the normal C C T A and it's a spec CT as well.

21:54

And if you look at the, at the r c curve here on the right side,

21:59

you can see again the blue line here on the top outperforms

22:04

even PET CT and SPECT C

22:09

in the detection of functionally, um,

22:14

relevant coronary stenosis.

22:17

So again, we have build up some,

22:21

quite some evidence and, um,

22:25

now there are still problems though, right?

22:29

We have only two approved products that are approved by the F D A and one is

22:34

using the computational fluid dynamics, which I explained earlier,

22:38

and that is performed off premise using a supercomputer.

22:42

So we need data transfer of your hospital and of course you're gonna pay for

22:45

that. So that's one difficulty that might,

22:51

uh, that one might face, uh, when trying to implement this.

22:56

But also the other, um, vendor is using,

23:01

it's not using computational fluid dynamics, but a machine learning algorithm,

23:06

but it's a proprietary algorithm,

23:08

so they're not disclosing that and it's trained on, uh,

23:11

that is trained on invasive F F R data and, um,

23:14

you still need your transfer your data to them,

23:17

and it's a subscription based model. So you,

23:21

you're still gonna pay quite heavily per patient.

23:26

And then, um, you need to transfer it in order for them to,

23:29

to do quality checks before they run their algorithms on it.

23:33

And if you hear cost, that is, that is sort of a, a difficult, um,

23:39

uh, topic because in from country Germany, we all,

23:44

we, everyone knows this evidence and everyone or should know, and, um,

23:49

we all know that we should use C T F F R,

23:51

but there's no reimbursement whatsoever for it.

23:54

So believe there's no clinical application for this at the moment

23:59

because costs are too high. Data transfer, um, regulations are too strict.

24:04

It's, uh, quite difficult.

24:07

But if you hear machine learning and you know,

24:10

one or two things about machine learning,

24:12

then that might spark your interest because you, you know,

24:16

machine learning algorithms typically really don't need that supercomputer.

24:20

So there might be a chance that they can be moved to like a virtual machine that

24:25

is run on premise.

24:26

So machine learning is something we should focus more on and uh,

24:31

I will dive into that right now. This is, uh,

24:35

this was the machine trial, um, which basically, um,

24:40

looked at the correlation and, um,

24:43

between computational fluid dynamic F C T F F R and machine learning F F R

24:49

and what it found it that both perform equally well.

24:53

If you look at the accuracy here on the bottom right, it's,

24:56

it's basically the same.

24:57

And if you look at negative and positive predictive values and specificity in

25:01

sensitivity,

25:01

you'll find basically the same values here and the

25:06

stenosis that were misclassified by C C T A were correctly

25:11

reclassified by both of those methods.

25:16

Similarly, um,

25:18

the next trial compared computational fluid dynamics versus

25:23

machine learning F F R versus invasive F F R.

25:28

And this is the point where we have basically gone full circle around to the,

25:32

to the first study that I highlighted. Um,

25:36

because you can see here that both machine learning F F R

25:41

and computation of fluid dynamic F F F R outperform normal C C

25:46

T A and Q C A. This just stresses, again,

25:50

anatomical imaging cannot determine functional, um,

25:55

functional stenosis, fun functionality of stenosis. So

26:01

we can use either machine learning or computational flow dynamic F F R,

26:06

but we cannot just do C C T A alone or Q C A alone.

26:13

All right, moving on. This one is an interesting one.

26:15

It'd say retrospective simulation studies on about a thousand patients.

26:20

And they looked at, um, what would happen if we would've added, um,

26:25

C T F of R to the C C T A, which is what you see here on the left. Um,

26:31

and you see the rate of I C A would've dropped significantly.

26:37

So the, that was the,

26:39

the thousand patients and we as was cut by more than half

26:44

the, if we would've used F F R C T. And similarly,

26:49

if we would apply it to those patients that had I C A, um,

26:54

then the rate of revascularization would go up dramatically because

26:59

then again, patients would've got,

27:03

would've gone to I C A with a higher pretest probability and we had would've

27:08

had less unnecessary icas.

27:13

So of course this is again some evidence, um,

27:17

that mounts up to guideline recommendations. And in the 2021,

27:23

uh, a h a recommendations and in the latest update of the cadra, uh,

27:27

recommendations, um,

27:30

it is recommended to perform C T F F R.

27:35

So now let's start looking into when to

27:39

really actually use C T F F R cataract's recommendations that were published

27:44

in J C C T here. Um,

27:47

recommend using it in stenosis between 50 and 90%.

27:50

That is basically the group cataracts, three to four, four a.

27:57

Similarly, the a h a recommendations, um,

28:00

recommended between 40 and 90% if it's an proximal or middle

28:05

coronary artery segment.

28:08

So what we can gather from this is it's not the patients that are fine anyway

28:12

that don't have a stenosis at all.

28:14

Those patients won't profit from C T F F R and it's also not the patients that

28:19

have an, uh, a total occlusion or a over 90% stenosis.

28:23

'cause that's something that's patients that need to go to I C A anyway.

28:28

So there's really no need for C T F R and it's not only that there's not,

28:33

there's no need for A C T F F R,

28:36

but it's also that the match between CT and F F R is

28:41

is not as good in these,

28:43

in these groups below 30 or 40% and higher than 90%.

28:49

So what is typically recommended, um,

28:53

and brings these two different recommendations in sort of

28:57

one is a anatomy based, um, sort of flow diagram.

29:03

So if we have low risk anatomy, so stenosis below 30%,

29:09

um, or high risk anatomy,

29:11

so left main stenosis or l i d stenosis, uh,

29:15

above 70% or three vessel coronary artery disease,

29:19

then we're talking high risk anatomy,

29:21

then we don't use C T F F R 'cause for these patients,

29:24

they either don't go to I c A at all or they go anyway and

29:29

in this intermediate risk anatomy or we are from 30 to about

29:34

69% stenosis or 70% stenosis.

29:38

Anywhere else in the left is where we use, uh, C T F F R.

29:45

And of course, if you look at the age a recommendation,

29:48

it gets a little bit trickier because it said proximal or middle coronary artery

29:53

segment that is due to the fact that of course,

29:56

if it's a very distal segment of that coronary artery,

29:59

that might not be revascularized anyway 'cause it's not a,

30:04

not a relevant target for revascularization. And then we don't need to, um,

30:09

need to do C T F R and recommended revascularization if it's not feasible

30:12

anyway. All right,

30:15

now that we've cleared when to F F R,

30:18

let's dive into how to F F R and they're

30:23

quite similar. And, um, most of the, uh,

30:27

or the two commercially available, um, uh,

30:31

systems will do this off-premise for you. But if you use a, uh,

30:35

on-premise solution, which are at the moment only scientifically usable,

30:40

not cleared for clinical service, um, you will have to do this on your own.

30:44

So what they do is they take a curvilinear projection of a

30:49

coronary artery and define a center line.

30:52

This is probably something you know from premier your everyday C C T A workup,

30:56

but then you have to define the lumen and be careful to exclude

31:02

plaques like these here. And based on that lumen,

31:05

you can then mark these stenosis in that lumen and that will then

31:11

yield this C T F F R model that we've seen earlier again as well.

31:17

And now look at these different colors and the, and the, uh,

31:22

the indicator here on the right side.

31:24

This will give you C T F F R for every point on

31:29

this, on this coronary artery. But in the same time,

31:34

there's different systems that will just, uh,

31:37

provide you with these kind of outputs,

31:41

which is you put in a curved linear projection again, and then it just models,

31:45

uh, coronary arteries for you and gives you one F F R value for,

31:50

for the entire coronary artery. So, which is best, well,

31:55

there's obviously evidence for that as well. There's,

31:59

these were the two approaches, the lowest CT C T F F R approach,

32:03

which just gives you one, uh, F F R value that has a high sensitivity,

32:08

but a low specificity. And if you look at the, um, uh,

32:13

this picture here on the, on the left side, you can, you can gather why.

32:18

'cause if you look at their normal right coronary artery here, um,

32:22

you can see that F F R volume decline naturally even in

32:26

healthy coronary arteries.

32:29

So if we just use this value right here,

32:33

we might be in a borderline F F R and might think, okay, this,

32:38

this coronary artery is, is diseased and we need to, uh,

32:42

refer for I C A, but that's not true.

32:44

So what we want to use is lesion specific C T F F R and is recommended

32:49

to measure one to two centimeters distal to that stenotic segment.

32:55

Then you have lesion specific ischemia.

32:58

That is what we wanna know because that then triggers revascularization

33:03

also. And really that that's something that makes sense if you remember how they

33:08

perform invasive F F R,

33:11

which is distal to proximal pressure, right?

33:15

So it's not all the way in the distal vessel, but it is right at the stenosis.

33:21

So this then yields higher diagnostic performance.

33:25

And this box right here on the bottom is the key takeaway.

33:28

Point eight or above 0.8 is normal C T F F R,

33:32

similar to what we've seen in invasive F F R.

33:35

Then for C T F F R, this is, uh, different.

33:40

There's a borderline, um, segment,

33:43

which is from 0.76 to 0.8.

33:46

And then everything equal to 0.75 or below will be abnormal F F

33:51

R and should be moved on to invasive coronary angiography.

33:57

So let's look at how this plays out in a, in the, in two cases. Uh,

34:02

the first one I have brought with me here is a 55 year old male patient,

34:07

but a three week history of exertional chest pain, no chest pain at rest,

34:11

and had an intermediate cardiovascular risk profile.

34:15

So right from the get go on the three D projection here,

34:19

you can appreciate some,

34:21

some nice plaques and right on the bifurcation and left anterior descending.

34:26

Um,

34:27

so let's look at the curvilinear projections and this should be a pretty

34:31

straightforward case for everyone. Um,

34:35

if you look at the left anterior descending,

34:36

you see a proximal stenosis that looks high level, uh, similarly in the,

34:41

in the left circumflex artery, you see a proximal stenosis right here.

34:46

And in the right coronary artery,

34:48

you need to go a little bit further down and then you can find it right here.

34:54

And then if we do computational fluid dynamics and uh,

34:58

reconstruct that, uh, coronary artery, these coronary arteries,

35:02

we can appreciate here in this very nice three D model for every,

35:08

um, lesion A F F R value below

35:12

0.75. So this is a proven three vessel disease.

35:18

Um, and this was, uh,

35:21

then confirmed by by I c A, um,

35:26

and then the also measured in the I C A.

35:29

So F F R was invasively measured and it was below 0.7 for all coronary arteries.

35:35

And the patient underwent subsequent cabbage as

35:39

indicated by the guidelines. So this was really confirmatory, right?

35:44

We probably would've moved this patient onto I C A anyway, based on these,

35:49

uh, highly relevant, um, stenosis just visually and not quantitatively,

35:54

but let's look at a different case where that might be not be as as

35:59

clear. This was a 59 year old male patient, stable chest pain,

36:03

a normal troponin levels and extensive cardiovascular risk profile.

36:09

So let's again look at the, at the curvilinear projections,

36:13

you can see that nice old plaque here,

36:15

right on the bifurcation of the first diagonal and the left anterior descending.

36:20

And this is difficult to quantify because where do you put your reference point?

36:25

Do you put it proximally? Do you only use one distally?

36:28

And this is not standardized either.

36:31

So this might be challenging in this case if you do it like that,

36:34

you come to a diameter stenosis of about 46%

36:39

intermediate stenosis, right?

36:42

So all guidelines recommend use C T F F R. So that's what we do.

36:46

And if we look at C T F F R two centimeters distally to that

36:51

plaque here, um,

36:53

we come out with a C T F F R of 0.76.

36:58

Now think back 0.76 is right in the borderline segment, right?

37:04

So we can really safely defer the patient, um,

37:09

and we can say, okay, this has to go to I C A,

37:13

this patient then went to I C a. And if you look at this right here,

37:17

I c A is, hmm,

37:19

it's not something you would definitely stand without doing F F R and I C A

37:24

anyway.

37:25

So of course what we need is invasive F F R and invasive F R

37:30

R was 0.77 and in invasive of r you

37:35

remember it's only the cutoff of 0.8 above or below.

37:38

And we said in frame two trial said treat everything that's below

37:43

0.8. So this patient then obviously got his stent and uh,

37:48

is moving on.

37:52

So this is a really nice case where we can reduce diagnostic ambiguity

37:57

and use it to guide patient's therapy.

38:02

But if we look at the curved linear projections,

38:06

you might come, you might think, okay,

38:09

well what happens if I have these image problems that we saw earlier?

38:13

And you're absolutely right, it's that classic garbage in, garbage out problem,

38:18

right? 'cause post-processing is always based on segmentation,

38:22

which is based on images.

38:25

So we cannot improve post-processing

38:29

capabilities endlessly without improving image quality.

38:33

'cause that will run into problems.

38:35

And depending on study you're looking at three,

38:37

3% to a third of exams are rejected for F F R analysis due to image quality

38:42

problems. And of course 33% is extreme.

38:47

Um, usually they're below 10% in the most of the studies. But

38:53

that will, that is something that has to be optimized as well.

38:57

And then machine lining, C T F F R, as I said,

39:00

there's actually no on-premise solution yet. And why is that? Well,

39:04

it's because they have,

39:06

the companies have proprietary algorithms and that usually leads to

39:11

disputes over patents and legal stuff like that.

39:17

However,

39:18

when we hopefully in the future have something that we can use on premise

39:23

and that is clinically, uh, marked, that will require your time, right?

39:28

Then you will have to do the post-processing and, and put in the effort.

39:33

So that is something you have to keep in mind as well.

39:38

And then there's sort of the outlook. What are other indications that all these,

39:42

all this evidence was for stable chest pain patients. What?

39:47

And it was for one stenosis. What about serial stenosis?

39:50

We know these patients usually don't have one more stenosis and then everything

39:54

else is fine. What about in the acute chest pain population? Right?

39:59

What about new scanning generations?

40:01

We have scanners that are able to provide ultra high resolution now, right?

40:06

And spectral data. What happens to F F R?

40:09

Let's take a quick outlook in that direction.

40:14

So this was a retrospective study that compared C C T A plus

40:19

C T F F R versus C T A C C T A alone.

40:23

And this was in a, um, in, in a chest pain setting.

40:28

So in an acute chest pain setting. And um,

40:30

they saw a high rate of feasibility in acute chest pain, um,

40:34

which you would not have expected.

40:37

'cause we talked about the rejection rate being the problem.

40:41

And in this study the rejection rate was around 2% actually.

40:44

So even in patients that are acutely sick and might have a higher, uh,

40:49

heart rate or had a higher incidence of, uh,

40:52

of arrhythmia at scan, they were still able to perform, uh,

40:57

C T F F R

40:59

and at 90 days there was no difference in in follow up between the C T

41:04

A group and the F F R CT group. Um, so again,

41:08

we can assume deferral of revascularization is safe for acute chest pain

41:14

patients, uh, with negative C T F F R,

41:17

but this needs to be validated in prospective confirmatory studies.

41:24

And this is something, um,

41:26

going into the direction of new scanner degenerations on the left here you can

41:30

see, um, the same coronary, uh,

41:34

from the same scanner and it's reconstructed in different slice thicknesses.

41:38

So on the very left you see 0.6 millimeters,

41:41

then 0.4 millimeters, and on the right side you can see 0.2 millimeters.

41:47

And if you just focus on that, on that slice here,

41:52

you can appreciate that the lumen actually seems to grow with increasing

41:56

spatial resolution. And that again is due to higher,

42:01

uh, higher spatial resolution leading to lower calcium lumen.

42:07

And now you might ask, okay, but if I contour this lumen,

42:11

I will have a broader lumen and than I have in this one, uh,

42:15

right here. So C T F R will be different, right?

42:19

And the answer is probably, but we don't know yet.

42:23

So that's work that is still has to be done. And we're,

42:26

what we're looking at right now is how does it do change anatomical

42:31

measurements,

42:31

but we're not sure yet how it will affect functional measurement as well.

42:37

And similarly, on the right side here,

42:39

you see spectral imaging is now readily available for coronary CT

42:44

scans.

42:45

And by modifying mono energetic levels with

42:50

different pre-specified stenosis, you can basically play around and,

42:54

and choose your grade of stenosis.

42:57

And that will also impact the lumen you define. And then again,

43:01

that will define your, uh, your your F F R calculation as well.

43:06

And you can see that in a patient, um, here on the bottom as well in this L A d,

43:11

you see the drastic calcium blooming with the same windowing and higher kav

43:15

levels you see a lot less stenosis.

43:19

So really what this leads to both higher resolution and spectral capabilities

43:24

will lead to fewer patients that have very,

43:29

very high stenosis, um,

43:31

and more patients that will have intermediate stenosis.

43:36

And that will then lead to a more request for C T F F

43:41

R. But we don't even know how C T F R works on these scanners.

43:44

So that's something that, that we have to keep in mind when,

43:47

when using it and when implementing it at our services.

43:54

So then we can conclude we have abundant validation evidence,

43:58

we have guideline recommendations to, to use it and um,

44:02

we have seen that it C T F F R uh,

44:05

reduces doubt in medical decision making.

44:07

It's non-invasive and it's specific and it does not matter if you use

44:12

computational fluid dynamic F F R or if you use machine learning F F R,

44:17

they're on par with,

44:18

with each other and they safely impact clinical patient management.

44:24

There are some,

44:26

some problems that remain or some challenges that we are still facing,

44:29

but it's a developing topic and I highly encourage you to, to try to,

44:35

um, implement it at your service and to gain some,

44:39

some,

44:41

some experience with it because I think it is something where we're moving to in

44:45

the future and that what we will use in the future a lot more than we are using

44:50

right now. And with that, um,

44:53

I'd like to conclude and I'm open to your questions. Thank you very much.

44:58

Ooh, thank you so much for sharing your lecture with us today, Dr. Hoffman. Um,

45:02

yes, at this time we will open the floor for any questions from our audience.

45:06

You may submit a question to Dr. Hoffman through the q and A feature

45:14

and hopefully we'll we'll get some questions going for you.

46:11

All right,

46:11

looks like we've had a few questions pop up in the q and a box for you, Dr.

46:15

Hoffman.

46:22

Let me see. Uh oh, there we go. Sorry. Okay, um,

46:28

let's start with the first one. Um, uh,

46:31

I would like to see how much it is studied in the emergency setting that

46:34

pertains to, um, to what we've seen in, uh, in the last,

46:39

in the outlook. Um, there is, uh,

46:43

few studies actually that studied it in a emergency setting and

46:48

studied it with, uh, for example, triple rule out CT as well.

46:51

And as you would expect, there's higher rates of rejections, um,

46:57

but it is still feasible,

46:58

but we are missing prospective validation evidence for that really.

47:03

So that's something that work that still has to be done.

47:06

We have retrospective evidence, but we need prospective evidence.

47:15

And then another question was, um, is there a formula to calculate flow? Well,

47:19

that is actually, um, as I said, it's uh, if you look it up,

47:23

it's a navier stokes equation, which is a highly complicated equation. Um,

47:28

and that takes into account, um, uh, basically the,

47:33

the flow, uh, and, and

47:37

simulates hyperemia and then, uh, normal, uh,

47:41

blood flow basically. Um, but it's nothing you can,

47:44

you can just calculate on your calculator, um, on the CT unfortunately,

47:49

which is basically why they use supercomputers for it.

47:57

And there's another question there, which is, uh, I think very,

48:00

very important and very interesting. How could foot on counting impact F F R,

48:03

especially with extended patients? And, um, that is, uh,

48:09

a very good question and that is exactly what you saw in the, uh,

48:12

in these outlook images. This is counting data, right? And this is what,

48:17

um, counting CT brings us.

48:19

It 0.2 millimeters and it is very high, uh,

48:23

resolution and that actually enables us to,

48:27

to image stent patients.

48:30

So usually for stents you would say, ah,

48:34

CCTA is probably not ideal, but in,

48:37

we have pre preliminary data that the high spatial resolution

48:43

actually lets you see stent struts and can,

48:44

you can actually see in stent re-stenosis. And again,

48:49

if you look at what are the cardiologists doing, um, they're,

48:53

they have been doing F F R for stents for a long time and they are actually

48:58

reevaluating, um, their, uh, intervention success with,

49:02

with invasive F F R.

49:05

So that is something that is entirely unvalidated yet,

49:08

but it's something that is a super interesting topic because

49:13

we are just learning how we can actually, uh,

49:17

sufficiently image send patients with C C T A and then applying F F R to that

49:21

is, is extremely interesting.

49:33

Another, uh, interesting question.

49:35

Do nitrates and basal violation influence C T F F R measurements?

49:40

Um, yes and no. Um, you could think so, um, because,

49:45

uh, if you give nitrates and baso violation,

49:48

that is something that we have said we need maximum hyperemia,

49:53

but you only, you get the,

49:57

the image and you images after your, your institutional protocol basically,

50:01

which in most centers I think will use, uh, nitrates, um,

50:06

in order to get a, a better, a valuable, uh, C C T A image.

50:11

And, but the hyperemia is simulated by the, uh,

50:16

by the equation. It is not, so it does not influence that equation.

50:22

So you do need nitrates usually in order to get the nice

50:27

image, which you can then process, but it does not influence the, the, uh,

50:32

equation or the, uh, measurements from that.

50:46

And, uh, then there,

50:47

there's one question pertaining to the cost of the software. Um,

50:50

that depends on, on the vendor, of course, um, you can estimate around, uh,

50:54

it's on a case basis, it's around, uh,

50:57

$1,000 per patient. And, um,

51:02

then as I talked about, um,

51:04

some countries are quite far ahead with reimbursement for that,

51:08

so that's not a problem. Um, other countries, like the one I'm living in,

51:13

um, there's no reimbursement at all.

51:15

And then a thousand dollars per PA per patient is, uh,

51:17

quite a lot and that will, um, basically render it not feasible to,

51:23

to perform F F R for every patient.

51:28

But then again, um, as I said, there's only two vendors right now and um,

51:33

that, uh,

51:36

makes it quite quite the small market and hopefully that will expand and prices

51:41

might drop a little.

51:46

And what can happen while doing C T F F R, um,

51:50

basically the same things as, uh, as their normal C C T A, right?

51:55

Because it's all post-processing and, um, we've,

51:58

we've got the evidence that it is safe to defer patients when C T F F R is

52:03

fine. The only thing that, um, you have to be careful is use it in the,

52:09

for the right indications, use it only in intermediate, um, stenosis. And,

52:14

um, then of course if there's, um,

52:17

if there's serial stenosis for example,

52:19

then these equations just don't take that into account properly.

52:24

So then that might, um, that might lead to,

52:28

to wrong measurements and that can then falsify, uh, your findings.

52:34

So if you move in between the boundaries that we've set by the,

52:39

um, by, uh, by the evidence, um,

52:43

it is safe to use C T F R. There's really nothing, um,

52:47

procedure related that can happen because it's just the normal C C T A.

52:59

Um, then there's one question if I, uh,

53:01

can refer to any available articles or data available, um, yes, sure. Um,

53:06

I encourage you to, to rewatch and scan every article. Um, um,

53:11

there's quite a lot of, uh, data there. Um, if you want to have like a,

53:15

a quick overview, um, I, I can recommend heavily recommend, um,

53:20

this, uh, this review here and radiographics,

53:25

um, which is really practical and,

53:28

and problem focused and, and gives you, uh,

53:33

um, gives you, um, an idea of how to use it.

53:38

Oh, you mean in regard to P C G? Um, well there's P C T and F F R,

53:43

there's really not much evidence out there that's a problem, right? Um,

53:47

actually there's basically no evidence, uh, out there that I know of. Um,

53:53

but, uh, for P ct, P C C T in general, um,

53:58

we are just learning the first few steps, um, of how that impacts,

54:03

um, coronary artery disease. And I think there's, um, there's a few centers,

54:07

uh, that do it and are really scientifically active. I'm happy to say that that,

54:12

uh, the mines University Medical Center is one of them. Um, then I think, uh,

54:17

Zurich is also one that is, uh, very, very active and investigating that.

54:22

And then the, uh, the M U S C,

54:24

so Medical University of Southern Carolina is also very active. Um,

54:28

so if you look at sort of, uh, for those authors, um,

54:32

you will probably, um, have a pretty good overview.

54:39

Imagery, construction parameters, um, are not very, um,

54:44

strict really. Um,

54:45

most vendors will accept your standard reconstruction parameters, um,

54:50

and they just, uh,

54:52

recommend or of course they have some quality checks in place,

54:56

so if you have large step artifacts or stuff like that, um,

54:59

they will reject the study and, and cannot perform F F R. Um,

55:04

but you don't have to, uh, do super high resolution or, uh,

55:09

spectral or anything like that you can do,

55:11

it's basically just your standard of care and you send it to, uh,

55:16

to the,

55:17

in to the software vendor and they will perform everything else.

55:22

Alright, Dr. Hoffman,

55:23

thank you so much for taking the time to share your expertise with us today and,

55:28

uh, thank you to all of you for participating in our noon conference.

55:32

You can access the recording of today's noon conference and all our previous

55:35

noon conferences by creating a free m r I online account.

55:40

Be sure to join us again next week on Wednesday,

55:43

September 13th at 12:00 PM Eastern featuring Dr.

55:46

Sun Mic for a noon conference entitled Update on Imaging in 2021

55:51

W H O C N S Tumor Classification.

55:54

You can register for this free lecture@monline.com and follow us on

55:59

social media for updates on future noon conferences.

56:02

Thanks again and have a great day.

56:04

Thank you everyone. Bye-bye.