Sponsored webinar | Adopting evidence-based AI technologies to advance your CTA program
morning good afternoon everyone Jonathan leik here from Vancouver I'm really honored to be uh moderating this uh this webinar brought to you by uh scct and of course hard flow I want to to start thank the pro program sponsors from heartflow for their support uh and also of course the great team atct as always uh perhaps why I'm stumbling a little bit is because I'm humbled to be moderating the session I uh have two of my esteemed colleagues and dear friends speaking with me uh Dr Ron blankstein and Dr Michael gallager uh really two uh exceptionally thoughtful clinician scientists uh who've taught me personally so much over the last 15 years really perhaps even 20 uh as it relates to the integration of CT understanding the possibilities of phenotyping artherosclerosis and really managing patients I think what we have for us today is not only exceptional imagers uh but tremendous uh uh tremendous clinicians as well so without further Ado I I'll kick it off and try to share some of my thoughts I'm going to focus Less on atherosclerosis and focus on other opportunities around uh um Lessons Learned as it relates to integration of artificial intelligence and machine learning in our clinical practice in particular I'm going to focus on ffrct as you know uh there are now many opportunities around anatomical segmentation of coronary atherosclerosis and we'll also hear from doctors Gallagher and and blankstein around quantitation of atherosclerosis uh and as well anatomical lesions but I'm going to focus on some of my personal lessons that I've learned from the integration of ffrct over the last seven years I do have disclosures as you know we run a CT core lab in Vancouver for which I take no compensation but I do give advice to Heart flow and circle cardiovascular Imaging so as I'm sure many of you know at the SEC I believe in 2010 we were first exposed to the opportunities of integrating artificial intelligence and machine learning as well as applying computational flu Dynamics to arresting CT to derive ffrct without a change in protocol without additional radiation and without administration of adenosine and importantly the technology we saw at that time was one that allowed for Point specific physiological adjudication of pressure lesion specific pressure focal uh pressure calculation distal to any individual lesion and it's that information that we've learned over the last 15 years that is fundamentally important to informing medical management informing the need for referral to the cath lab and enriching the cath lab referral such that patients who go to the cathb are more likely to be intervened upon when we look at the accuracy of ffrct I share this slide and you may say why are you sharing a slide five years ago from five years uh from five years ago and the Really the reason I am is really to share with with you what I've learned over the last five years if I would have have shared this slide five years ago I would have focused only on accuracy the accuracy of ffr CT compared to invasive ffr highlighting that if the invasive ffr was 081 or above or 08 Z or below ffrct would be on the correct side of the cut Point 87 times out of 100 but what I've learned over the last number of years is that accuracy is only one modest measure of diagnostic performance as it relates to what the clinician and what the interventionalist needs to know when you look at the guidelines the guidelines suggest that we should consider using ffrct in a 40 to 90% stenosis in a patient and a vessel we would consider for revascularization to inform the decision around cath referral but when we're informing that decision we need to really understand what the interventionalist needs to know she doesn't want to know if the lad is likely normal or abnormal in a binary sense she actually fundamentally needs to know whether or not the lesion itself is causing abnormal pressure how abnormal the pressure is and how the pressure falls off in relationship to the focal lesion and that is why unique to ffrct is the ability to compare the absolute value in a quantitative and statistical fashion which in the setting of the NXT trial was excellent point A2 that means the absolute value agrees in an excellent fashion with the invasive measure so we're not telling the clinician that the lad is normal or abnormal uh independent of where or the absolute value but rather really isolating the lesion and telling the clinician that the lesion itself is causing abnormal pressure loss and this is the absolute calculated pressure and this is the pattern of pressure loss across the lesion now that may be a lot of words that seem complicated but at the end of the day it's quite a simple concept this concept that physiology has no utility utility in as a binary tool if we look on the left you could see you have an lad system where the pressure falls off in the distal diagonal the lad system is abnormal that's a true statement but there's nothing focally to treat with mechanical intervention the lesion specific physiology is negative there's no focal drop across it and rather you just have slow dissipation of pressure across the length of the L and the Dual diagonal on the other hand on the right you have focal lesions and each lesion itself is causing focal pressure loss with the absolute values distal to the focal lesions being abnormal both of these models are abnormal but the one casee on the right is a model that will derive benefit from referral to the cathb for planning complex PCI or bypass and the case on the left needs medical therapy we've learned this over uh now the last seven years this is some of our original work highlighting the importance of isolating the lesion in informing the decision around cath referral you identify a stenosis Dr blankstein has read it and the question is will this patient derive benefit from cath referral or not and to answer that question question he doesn't want to know if the lad is likely normal or not he really needs to know whether that lesion itself is causing abnormal pressure loss and in this case you can see the lesion itself is negative sure the distal vessel drops below 08 but there's nothing focally to treat with mechanical intervention and we now have seven-year follow-up data that will be presented for example at TCT that highlights that managing patients based on the lesion specific value is fundamentally helpful in informing the decision around Cath and also informing decisions around Downstream medical management this has been shown of course not just observationally but also in randomized Trials the precise trial looking at the lesion specific value informed decisions around cath referral in a fashion that drastically and statistically significantly reduced the rates of cath without obstructed Disease by 70% while improving the PCI to IC ratio from 28 to 72% in this trial taken to an stre again both of these LEDs are abnormal this is a true statement neither of them of I can assure you none of these models are those of P of people playing in the NFL but the case on the left you see very slow dissipation of pressure without a focal drop that patient needs medical management despite the fact they have diffuse atherosclerosis and a stiffened or hardened artery there's nothing focally to mechanically intervene whereas the case on the right of course has focal pressure loss with abnormal translational pressure loss and a markedly abnormal lesion specific value of 66 finally I just wanted to close on one of the things that we've learned as it relates the pattern of pressure loss you know if we were to take a a time capsule or go back in time to 2009 with the original randomized trials in the invasive space around ffr we were we learned that physiology as we've seen now repeatedly should be used to or should be considered to guide decisions around revascularization of course now we know that invasive physiology uh is in the guidelines for the chronic coronary syndrome management as well as for revascularization as class one indication but what I've learned from our Interventional colleagues over the last decade is that it's not not enough to Simply say again that the the lad is normal or not we need to interrogate not only the depth of the pressure loss but also the pattern of the pressure loss as shown here from Dr Colette they've dropped a wire in the lad and they're pulling the wire back interrogating the phenotype of pressure loss across Ross this the length of the lad all three of these arteries in a binary sense are abnormal they drop below 08 in the distal vessel but on the right the pressure falls off incredibly slowly across the length of the vessel on the other hand on the left you could see that there's a focal drop of pressure across a focal lesion this really will lend itself to Mechanical intervention and again on the right this will lend itself this will not lend itself to Mechanical intervention this is actually from uh the people PPG Global study published recently by Carlos and his colleagues in in circulation and again you can see that both of these invasive ffars on the top row are abnormal in fact the case on the on the right is significantly worse than the case on the left but when you look at mechanical intervention you can appreciate that that focal pressure loss as shown here on the top left really lends itself to Mechanical intervention and the post PCI ffr is significantly improved on the other hand on the right you could see that the pressure loss is incredibly slow and that despite the severely abnormal physiology def long segment mechanical intervention does not meaningfully changed the pressure with a post PCI FF 5.66 so over the last seven years we've learned that this is fundamentally important and we use and we and we keep this in mind as it relates to interrogating the vessels as we evaluate them prior to decisions around referral to the cath lab all four of these LEDs are abnormal these are all normal physiological LEDs with pressure loss in the distal vessel but on the right you see very focal phenotypes of pressure loss really lending themselves to Mechanical intervention whereas on the left you see this diffuse pattern of pressure loss again highlighting an abnormal LED but an LED that is unlikely to derive benefit from cath referral and PCI uniquely ascertainable or able to be discriminated non-invasively using ffrct so I will conclude my my few words words will pivot to atherosclerosis but I just wanted to close with lessons that I've learned over the last seven years that essentially per vessel binary physiology has almost zero clinical utility in my practice I can predict that the lad is likely to have abnormal physiology in the distal vessel when I see diffus disease in an anatomical lesion my real question is not is the lad or the right coronary normal or not my real question is will this patient derive benefit from mechanical intervention of that focal Le and in order to do that I need to localize ffrct disle to the leion I need to interrogate the depth of the pressure loss and I need to characterize the pattern of the pressure loss all three of these elements are fundamentally important in guiding ICA referral and PCI planning as we've seen now observationally with up to seven-year follow-up and in again two randomized prospective trials so I'll stop there and I will pivot over to a dear friend Professor blankstein who comes to us of course from from Harvard from from the Mass from the Brigham and Women's Hospital past president of the SEC current chair of the ACC cardiovascular Imaging Council but most importantly an incredible friend uh and a very dedicated clinician scientists who's taught me and so many of us uh about cardiac c as well as prevention so without further Ado Rowan please well Jonathan thank you for the very uh kind introduction and uh always an honor to follow you and follow an outstanding talk I'm going to talk more about plaa analysis we'll kind of shift a little bit to plaque analysis and one of the really exciting developments in plaque analysis is what we now called quantitative coronary plaque analysis or qcpa and I'll talk about the accuracy of that and I'll show you uh an example so as as a background one of the one of several studies that really kind of set the the the tone and showed us the important prognostic value of quantifying plaque was the Scott heart study and shown on this slide here is a very important study that was led by our colleague Dr Michelle Williams uh where plaque was uh Quantified these are individuals in the Scott hard trial which of course is a trial of patients with uh with stable chest pain and what we can see on the top uh slide here is the prognostic value of total plaque volume the overall amount of plaque that individuals have and in this study patients who have the highest plaque volume have a substantially higher rate of myocardial infarction and importantly this is increment or Beyond risk factors calcium score or whether patients have stenosis and that's a very important point to show that plaque analysis not just tells us who's at higher risk but to show that it's actually incremental to other things we typically measure on CT we then can characterize plaque and there's different ways to do so but one of the most common ways is to look at the type of plaque by looking at the houndsfield unit and dividing a plaque into non-calcified and calcified and then the non-calcified plaque can be further divided into into uh different types uh with low attenuation plaque being specifically plaque that has a houndsfield unit less than 30 and in this study here uh uh Dr Williams and our colleagues showed that low attenuation plaque have the strongest association with incident mi in the in the Scott har study uh specifically when individuals had more than 4% uh burden of low attenuation plaque uh they had the highest risk so so important study showing us that that plaque really matters in qu ifying it makes a difference when it comes to risk assessment but of course the big question is how accurate is plaque modification that just kind of tells us who is at risk but how does it correlate with other um techniques and probably the best one and what would be the reference standard would be invasive technique such as ivis and this is what's been done in the next two studies that I will show you the first one here is the Miami study so this is out of uh William Bowmont it's a single Center uh study where individual who had a stemi uh were recruited into the study they had two invasive studies but in addition to that they had a cornery CTA and that allowed the investigators the opportunity to quantify plaque and CTA and see how it correlates with invasive measur specifically with IIs and uh the results are shown over here and on the left uh just to kind of walk you through this figure here we have on the Y AIS the total plaque volume is determined by ivis on the x-axis the total plaque volume is determined by coronary CTA and what you can see is a very strong correlation here an R value of 0.92 so 1.0 would be absolutely perfect nothing is almost nothing is ever perfect but 0.92 is a very strong correlation uh in impressive data on the right side of this uh slide further dividing this into different types of plaque so uh not just a total plaque volume that has a correlation of 0.92 but calcified plaque
and non-calcified plaque also was uh was strong correlation so this is a single Center uh study but if we now move beyond that uh there is now a multi- center study and I think an tremendously important study in the field and that's the reveal plaque study this is a study that was uh conducted at 15 sites um where plaque was actually measured both by ivis and by CT and this was a prospective comparison it was done very well cor laabs did this so there was a COR laab that measured plaque was IIs and of course also was CT and on the left side here you can see the correlation here 0.91 so very similar to the Miami results but now shown in a multi- center uh study showing very strong correlation when we look at plaque volume and specifically there was also good correlation when looking at the subtypes looking at calcified plaque volume and non-calcified uh plaque volume uh in this study I would note the the the third correlation that was done was low attenuation plaque and probably as a expected lower attenuation plaque a little bit of a lower correlation than than calcified and non-calcified plaque and I think this is expected because the different threshold for low attenuation plaque I think uh vary and there's uh I think technically it's a more difficult measure to get so how does this uh work uh clinically how would we use uh plot quantification I wanted to show you uh one case which I thought was was uh interesting and instructive from my own uh institution uh this is a cornery CTA of 32-year-old male who uh has hyper lipidemia um yes LDL is high 150 milligrams per decil but more impressive is his elevated lipoprotein a which is 230 nanom per liter um to kind of give you a sense of what that is uh the current ongoing clinical trials that evaluate therapies for lowering lipoprotein a are using a threshold of 200 or 175 to get into those trials he was also a smoker and he was referred for a coronary CTA after he went to another hospital and uh had what was ultimately determined as a non-st elevation by a cordial in Fortune he had some uh fairly non-specific chest discomfort but he was found to have a mild elevation in his troponin so he was diagnosed was a non-st elevation mordial infarction it was actually attributed at the time to the use of uh marijuana because that that was also a complicating Factor so he is sent for a cornery CTA and what we can see see on this uh uh slide over here is the coronary CTA results and specifically if we look uh at the left at the right corner artery the lad and circumflex we can see that he doesn't have any significant obstructive disease really minimal plaque our eyes are always drawn to the calcifications but if you look carefully you may see well perhaps there's some non-calcified plaque U and overall I think in most of our patients this would be a small amount of plaque perhaps when you're just kind of looking at these quickly but of course because he's young I think it catches our attention a 32-year-old really shouldn't have any uh plaque ideally and when we look uh at the um images we actually start to see some more plaque but in this case um initially we thought he had plaque I don't think we were as impressed by the amount until we actually did a quantitative plaque analysis and this was done uh by by the heart flow tool that we're discussing today and what we can see here when we perform that this is the the ultimate readout is we can see for each vessel l main LED circumflex RCA the total plaque volume in in cubic millimeters and you can see that on the bottom and then for each vessel and for the total that's then divided into calcified plaque non-calcified plaque and low attenuation plaque which again is a subtype of the non-calcified plaque in his plaque volume 824 cubic millimeters that's a lot of plaque for anyone but particularly for someone in his uh 30s um now how do I know that's a lot of plaque well of course when you start to quantify plaque you kind of get a sense to of these numbers but if you're not uh sure I think one area that's very useful is a nomogram and uh Jonathan leic actually uh helped develop this and on this nomogram here we can see the percentiles for these different curves across different ages now of course he's less than 40 so he's off the chart and if I put this red star for his amount to plaque which is 800 cubic millim we can see he's like way above the 90th percental in fact if you are young even having a more than uh 100 cubic millimeters of plaque would put you above the 90th percentile he of course has you know tenfold the amount that would put him at the 90th percentile so extensive amount of plaque but not only is the amount of plaque is high I think the other uh uh very interesting point here is that 90% of this plaque is non-calcified um we are always interested in who are the patients who are more likely to have non-calcified plaque and probably the reason why this catches a lot of attention is because this is a type of plaque that's more modifiable it's a type of plaque that perhaps we can uh regress was our therapy so this is really the target of most of our therapies as the non-calcified plaque um and we know that younger patients specifically if they have significant risk like if they have diabetes or familial hyper cholesterol Aria or in this case elevated lipoprotein a um systemic inflammatory diseases there's various conditions that we are learning more and more that they are are uh risk factors for having more non-calcified uh plaque the other aspect here is the low attenuation uh uh plaque so again a subtype of non-calcified plaque and in this case this represents 10% of the plaque so that is a that is a furly large amount of non-calcified plaque so on this slide here I want to show you a little bit the the AI uh the interactive tool that we use to look at plaque so specifically if I kind of walk you through this video here there's a toggle switch where you can go between ffr CT and plaque if we're now in the plaque modules we can go up and down each vessel so right now the LED is selected and as we go up and down and you look at the right side of the screen the plaque is colorcoded where in pink and now we're shifting to the RCA but same colors here where blue is calcified the pink is non-calcified and the purple is the low attenuation plaque and we can um in a very interactive fashion you can pick each artery go up and down the artery and see exact where the plaque is it's interesting here that you actually see circumferential plaque and it's not common that you really see plaque going all the way around an artery um and of course when I I I looked at this I said wow this is a lot of plaque and we go back when we see this and we say let's look at the images can our eye actually detect this plaque and this is really one of the powers of quantification is seeing things that our eye can see but sometimes not as well so if we actually go back and look at the images in this specific uh cas case here on the top is the LED on the bottom is the RCA we can actually see this um you know low attenuation plaque it's a little fuzzy looking but we can see as we're kind of doing the equivalent of a pullback through the LED on this image here we can actually see circumferential plaque uh in some of these portions here and we see extensive non-calcified plaque same as the right corner artery when we kind of go through and look at this vessel carefully we see an extensive amount of non-calcified plaque with very nice correlation between where we see plaque and where it was actually Quantified so uh 10% low attenuation plaque does that mean anything and of course I already hinted that that signifies a a higher risk but one of the big questions in cardiology and I'm going to use lipoprotein a as a as a case Point here is can plaque Imaging improve risk assessment and specifically in patients with lipoprotein a this is a very hot area now because there therapies on the horizon but when we talk about lipoprotein we eval the literature as summarized in in this figure here put together by Arthur shich we see that elevated lipoprotein a Associated not just with more plaque but faster plaque progression more high-risk plaque It's associated with more abnormal lesions by ffrct more obstructive disease a higher calcium score uh and also low attenuation plaque so basically all the bad things that we can see with CT uh have been associated with elevated lipoprotein a so uh how do we put this all together how do we interpret this case so our interpretation in the report is as follows there's extensive amount of coronary plaque the total amount of plaque is 824 cubic millimeters we indicated in parentheses that this is greater than 99 percentile adjusted for age and sex with 88% of the plaque being non-calcified in the low attenuation plaque uh represents 10% the recommendation here follows the cadra guideline uh and specifically this is a cadrs 2 two means mild stenosis of course we don't get too excited about mild stenosis on CT but this is a P4 P4 is the highest uh category of plaque as a reminder in CAD RADS the the pcore goes from P1 to P4 and when it's P4 our recommendation is referral for outpatient follow-up for aggressive risk factor modification and preventive far pharmacotherapy we now put this uh by the way in every report the recommendation so this is the utility of plaque Imaging is that we can get a more precise estimation of the amount of plaque and this directly ties to our recommendations so our take on point is that quantitative coronary plaque analysis can provide a quantitative measure of the amount in type of coronary plaque and by type I mean the specific morphology that we talked about which is non-calcified or low attenuation plaque or calcified plaque the information can enhance cor CT interpret ation refine patient risk assessment and also help identify individuals who are most likely to benefit from aggressive preventive therapies so I will uh end it there and turn it back to uh Jonathan thank you so much Ron what a fantastic talk we're going to have lots of time for conversation I really look forward to it I was taking some notes and formulating my own questions but for those of you listening in please enter your questions in the chat we'll get to them after Dr Gallagher's uh final presentation thank you again Ron so it's a real privilege again to welcome Dr Gallagher he's a program director of the cardiovascular disease fellowship and the director of advanced cardiac imaging at corwell Health Formerly Known at as Bowmont hospital he's a professor at the Oakland University William Bowman School of Medicine uh and like Ron a dear old friend of mine uh what what you'll see from Dr Gallagher is he's really not just a a master CT uh interpreting physician but also a master clinician he's taught me so much about the integration of CT and how he uses CT to manage patients and I look forward to your presentation Michael thank you Jonathan uh thank you uh Ron and Jonathan and in particular the Society of ceser CT and heartflow for giving us the opportunity to really share some exciting data and cases with you all today and so my task today is just that is to follow the really great lead-ins from Dr liick and Dr blankstein um and to share another case that highlights some of the advantages of uh plaque assessment so uh I do spend some time working as a mentor or consultant for folks who are starting up uh ffrct programs so my case is a 52-year-old physician with a past medical history of hyper lipidemia and hypothyroidism his father underwent bypass at 62 years of age he was on a low dose of a low intensity Statin um I saw him in my office for intermittent chest pain and the pain was uh typically at rest more recently uh associated with a left arm ache after playing hockey and sometimes while playing hockey of note he had a stress echo cardiogram a few years ago that was negative his blood pressure was mildly elevated his total cholesterol is 255 his LDL was 139 and his hemoglobin A1c was 5.6% you can see his EKG fairly unremarkable shown here so a patient like this who comes to me with chest pain in my office or for that matter to the ER with chest pain we typically lean on a CT first uh strategy and that's what we did in this particular case you can see in the bottom uh his right coronary artery has minor calcified plaque in the proximal and mid to distal segments and more interestingly in the L in this eloc crany view you can see fairly extensive non-calcified plaque followed by calcified plaque as shown in the cross-sectional regions this was really felt to represent a greater than 70% stenosis in the proximal to mid LAD and this was sent off for ffrct and as you can see here to follow Jonathan's lead in the translesional gradient was really quite unimpressive from 0.99 to 0.92 Across the lesion the the mid to distal ffrct value is 0.9 so these findings suggested the lack of lesion specific esema so in summary we had a patient with an lad proximal to Mid segment calcified as well as non-calcified black with a greater than 70% stenosis or a cads cad Red's 4A designation but an ffrct of .9 that
suggested the lack of lesion specific es keemia and the plaque burden and the plaque staging system was a P2 to a P3 so with that we treated the patient U medically and uh what I want to share with you is if we look for a minute at what Dr blankstein shared with you on the farle model of our interactive plaque assessment tool we sent this data offer interactive plaque analysis you can see the 3D model to view the ffrct and the plaque the left as well as the total plaque volume on the left in the middle we have a straightened CPR that allows us as Dr blankstein shared with you to track along the vessel to look at the ffrct values as well as the extent and type of a plaque and on the right in this particular case yellow is designated as non-calcified plaque and blue as calcified plaque on the far right portion of the screen so if we put this into action you can see that the ffrct value is .92 distal to the stenosis and if we then track up and down the vessel and I bring my cursor back to the left Main and proximal LED keep your eyes focused on the right and look at the degree of yellow non-calcified plaque as shown here followed by Blue calcified plaque fairly extensive non-calcified and calcified plaque throughout the proximal to mid LAD with a plaque score alone in the LED of 419 cubic millimeters so this is our interactive plaque assessment coupled together with our ffrct assessment a negative ffrct but fairly extensive plaque burden so if we take a closer look at that as you saw before Dr blankstein we can actually summarize the overall plaque score and this patient had a plaque score of 619 cubic millimeters but again similar to the last case heavily uh associated with non-calcified plaque with a non-calcified plaque score of 526 cubic millimeters and if you look more closely the colors the design you can see I've shown you the yellow in the longitudinal vessel the straightened CPR in the bottom correlates nicely with the short axis images the blue calcification as shown here it's a nice color depiction of what we're looking at both in Long axis and in short axis as well as a summary of the degree of plaque so I had to put this all together I saw the patient in my office after the cardiac CT and the ffr CT and the plaque assessment and the decision was made to maximize medical therapy with a diagnosis of non-cardiac chest pain there were several atypical features associated with the pain as well uh and so there was no plan for invasive cornea angiography however of course we needed to maximize his medical therapy and so we worked on that however over the next six weeks or so the patient had intermittent continued chest pain and ultimately presented to the emergency department um and the decision was made by the admitting team to proceed with invasive cornate angio graphy so if we take a look at what the invasive angiography showed in the left you can see a 30 to 40% LED stenosis and in the middle you can see the hyperemic invasive ffr assessment after a Denine was 0.88 distal to the stenosis which correlated very nicely with our non-invasive surrogate ffrct so both of these demonstrated the lack of lesion specific esia so this patient was diagnosed with non-cardiac chest pain no PCI was performed but as you can see in the bottom right we did take the opportunity to put a wire down the artery and take a look at the artery not only for invasive ffar but also for intravascular ultrasound and you can see this run pullback from the distal vessel to the proximal vessel on the bottom right and we initiated medical therapy on this patient but what I want to share with you is that when we look at that pullback and we look at the degree of plaque shown outlined in blue here the calcified plaque you can see on the ivis the calcification in the proximal to mid LAD with lots of shadowing to the right of the calcification correlates beautifully with the uh plaque assessment and then again all the non-calcified plaque shown by the yellow arrows nicely outlined again on the IIs in the panels above and on the plaque assessment by heart flow in the panels in the middle so a nice correlation of the type the characterization and the Quant quantity of plaque with ivis compared with CT and it really does bring home some of the data that Dr blankstein shared with you about the revealed plaque study because in this study as you saw there was a as well as with Miami there was an excellent correlation with the CT quantification of the characterization and the extent of atherosclerosis demonstrated by cardiac CT as compared to the reference standard of Ibis and so our case was a very nice uh depiction of this and so let's bring this back home and talk about how do we as clinicians integrate all this data into our practice and how do we manage the patient and we need to put this together into a report and Dr blankstein commented on this a bit but I would share with you is that you've seen that there's an age and sex specific nomogram as Dr blankstein shared that we can now categorize my patient in reference standard to thousands of patients um with a similar age and gender and we can plug his age type of plaque into a plaque calculator and determine that this patient is at the 84th percentile for the overall degree of total plaque with a plaque square of 619 cubic millimeters so now we have some objective data and a percentile but as Dr blankstein also mentioned it's so important in the cad RADS 2.0 system to designate not only the degree of stenosis severity but also the overall amount of coronary plaque and initially it's can be very difficult because this is a very subject assessment and this patient was felt to be between a P2 and a P3 it's very Visual and subjective about the number of vessels or the extent with a visual assessment of the degree of plaque and now what we're able to do is create staging systems and this is taken from a registry that is looking at 20,000 patients and they're stratifying patients according to their total plaque volume and they're characterizing the mild plaque volume with a plaque score of one to 100 moderate 100 to 250 severe 250 to 750 cubic millimeters and extensive greater than 750 cubic millimeters and so what we've started to adopt is to actually classify this is our P1 through P4 staging systems so now we have a relatively accurate quantitative and reproducible way in which we can characterize our overall plaque burden and so now our reports will not only show the cad RADS 4A in this particular instance but also can share the overall amount of plaque as based upon a plaque characterization tool with objective data and I think we're really moving towards personalized risk prediction with these sorts of scoring systems so um let's get back to my patient what did we do well we treated him for his non-cardiac hockey related muscular skeletal pain he saw anesthesiologist for injections of his thoracic pain he was treated with a maximum dose of a Statin we added a zami his LDL was still above 70 so we added a PCS K9 inhibitor he was initially reluctant to treat his blood pressure aggressively but after seeing the degree of black in his arteries he was agreeable to maximizing anti-hypertensives we added aspirin focused on exercise and diet and we really started to think and talk about early treatment of borderline diabetes as his A1C started to increase on subsequent visits so with that I want to bring that home to some take-home points and I think with this case we were able to share the importance of ffrct assessment to help guide decision-making the importance of a physiologic uh map on top of our anatomy a greater than 70% stenosis but a negative ffrct very important for your clinical care we talked about incorporation of plaque assessment tools to quantitate the degree of atherosclerosis we we really verified some of the data from reveal plaque in Miami about the accuracy of quantification and we are able to calculate a patient level plag volume integrate all that data into a report and we can use that data to better inform our management decisions so with that I'd like to thank you all and maybe turn this back to Jonathan for a uh a discussion so thank you all thank you that was really a great overview um as I as I thought not only of the science and the data but both of you sharing your clinical experience which is so powerful not just from a visual interpretation and and reporting of the CT of course but but how you would manage these patients and that's incredibly uh helpful powerful for me personally I know and I'm sure for so many others uh on the zoom so uh let let's uh welcome uh Dr blankstein as well back in and let's have a conversation I have a few question there's a question in the chat that we'll also address but before I do uh I think both of you highlighted the fundamental importance of uh tool validation you know we have a lot of tools in imaging I'm a radiologist by training as you know there are many tools that are out there for lung nodal detection and and stroke detection and blood intracranial hemorrhage detection and one of the things that I'd like to teach our fellows and residents about is is the fundamental importance of of the hierarchical evidence required to integrate a tool you know the idea that you need to show that a tool is first reproducible and accurate and then of course you can explore the risk prediction and clinical utility and then the prospective integration but you both highlighted the the ivis correlative data of Miami which comes from your Center obviously Michael um and then the reveal plaque study led by Professor nula the largest International multicenter ivis correlative study and maybe as clinicians you can tell me you know many people will say listen I visually see the plaque I know when there's a lot of plaque but but why is accuracy important to you uh as it relates to uh understanding a tool before you integrate it into your practice I don't know Ron maybe let's give Michael a break let's uh go to you why do you think those data are important why did you share that in particular yeah well I I actually think in imaging people are kind of obsessed was quantification and numbers recently it's like a trend you know in uh in MRI we never had numbers now we have all this mapping we give all these numbers uh and then uh cardiac pet we the new thing is quantifying blood flow um in Echo there's various quantifications and strain and I I think the problem is that often we have a tool and it gives us a number and sometimes there's a lot of um uh overemphasis on some of the numbers you know people report I we we report an EF to the hundreds we say the EF is 47.5
2% and I say is it really that accurate so before we start reporting numbers it's really important to know what those numbers mean before we start putting numbers in report what you know how have they been validated what is the accuracy how is it compared and I think it's really important for imagers because imagers whether you're cardiologist or radiologist if you put that in the report you're referring physician to ask you well actually how what does this number mean and how is it validated where did it came from uh and then of course I'll ask you what do I do about it U but I I I think it's a really important thing is we're using more and more numbers uh I kind of urge everyone to always question where those numbers come from and how are they validated I think it's essential yeah that's that's a really good point and I you know when I think about the utility of plaque tools I know that I've tuned my eye to to evaluate coronary disease obviously over the last 20 years and I know when I read a CT as negative those patients generally do well and so I I think you highlight a very important Point particularly around the identification of modest amounts of plaque uh if a tool is going to tell me there's plaque there I'd like to know how well they agree with the intravascular gold standard at identifying or characterizing modest amounts of plaque because you know it'll it'll be perhaps uh contradicting my visual interpretation and so I really have to have some have some confidence in that uh and you know maybe Michael well I'll ask you then there's a question from the audience that I think is important and there's probably not a a definitive answer at this point but it highlights an important outstanding question for the field is you know how do the different AI uh derived total plaque volumes correlate and I assume the individual is asking you know across the different vendors and and um you know your thoughts about what sort of science we would need to do or or what would need to be done to try to answer this question or or if you don't think it's an important question you can obviously share that as well no I think it's a good question and and I guess the first um thing to reflect back a bit on what Ron shared is that you know I I do think this um you know reproducibility and accuracy is very even you and I and others when we have intra and interobserver variability and even something as simple of as a P0 through P4 staging system I think we often may even disagree on that and so I very much welome welcome this uh quantification to integrate into a very well accepted CAD RADS 2.0 reporting tool and I think it lends itself extremely well into uh a recognized and adopted uh uh uh Reporting System but I think as far as the question is um you know uh different vendors have looked at the degree of atherosclerosis in correlation with intravascular ultrasound and I I think back a little bit to No Doubt what Ron's quite familiar with with strain Imaging you know strain Imaging um is very vendor specific and there's some differences and subtleties from one to the other in terms of absolute values of strain Imaging when you assess strain performance of of heart function and and that's okay we've learned to deal with that it's not a a deal breaker by any means and it's something that we recognize as a scientific community and we move forward and we use strain commonly so I think we may have some of these um lessons to learn as it relates to atic plaque characterization and that's okay uh but I think we need to research it study it recognize it and know how to incorporate it into patient care great great comment great answer to the very challenging and interesting question I mean we have all sorts of issues I know from the scct uh there's been a real effort leadership uh I think that was under under your uh guidance Ron but leadership from uh doors Neiman and Garcia around quantitative uh plaque in in and CT and sort of a guidance document highlighting the many opportunities but also the challenges I mean what I'm very I'm very excited about plaque quantification but I'm also very aware of of our history in cardiac CT when we ran too quickly with cardiac CT in 2005 and 2006 and 2007 and we we didn't necessarily have the evidence or the data and we were uh really uh perhaps overly ambitious and excited uh and we we we ate Humble Pie so I think that you know the the thought thoughtful integration of plaque uh is going to be very important you know acquiring the data appropriately using CT in a thoughtful way if we're doing uh tracking treatment response we obviously need to visually inspect those CT data sets as well so let me let me uh pivot back to you Ron um you know I was very kind of both of you to highlight George Seamus's work you know around the nomograms obviously these are referred patients it's not asymptomatic so it wasn't intended to be though but you you highlighted the importance of standing the prematurity of plaque or the percentile of plaque and uh maybe you can share some of your thoughts Ron obviously now two decades of thought leadership and calcium scoring you know many people will tell me you know it's just the burden of plaque and and obviously that's true at two years if you have a truckload of plaque uh you know even if and you have a very modest amount of plaque even if it's premature the person with a lot of plaque is more likely to have an event but clearly the prematurity of plaque in a young patient may not may not be relevant at one year or two years but I'm sure as a preventive cardiologist that's that's quite important to you as it relates to their their 10year lifetime risk maybe you can shed some light on those two elements yeah it's a it's a great question Jonathan really it's like what what's more people ask this a lot what's more important the percentile or the absolute amount of plaque and the the reality is they're they're both important so in the in the short term whether it's a calcium score or CT the overall amount of plaque you have I think is going to be the strongest predictor of risk if you have a lot of plaque you're always going to have higher risk um but percentile helps us identify folks who are kind of on the extremes better so young individuals may not have a lot of plaque maybe they only have you know 80 cubic millimeters of plaque but for their young age they may be over the 90th percentile um and that's a that's a reflection that's helpful both for relative risk so I would tell that young person relative to other uh males your age and your race you are at a much higher risk you have more plaqu so it's it's a measure of relative risk but it also ties into long-term risk uh somewhat analogous to the fact that many of our risk calculators and in uh prevention really look at risk in the next 10 years and now there's an effort to look at lifetime risk uh it's the same thing if you have plaque at an early age uh I'm going to be I'm not going to be concerned about your risk of having an MI next year but I'm going to be very concerned about your lifetime risk of cardiovascular disease because I'll say if you're starting to have this much plaque at this age you're on the wrong trajectory so so really important uh concept we have to look at both uh the absolute risk at least in the short term is going to be the absolute amount of a plaque but the relative uh risk and the long-term risk I think is going to be more informative to look at the percentile thank you that's um yeah it's really exciting and I think particularly if you're thinking prevention and understanding treatment response it's it's going to be highly important as you point out really great question to bring us home perhaps from both of you um interesting question from the audience are there any studies uh investigating whether or not delineating plaques based on morphology using uh AI to predict which plaques May rupture and guide guide premature stenting well I guess it wouldn't be premature but maybe I guess it would be preventative stenting um I'll I'll get both of your thoughts I'll share uh for those in the audience recently Professor coup again published a a very interesting study called Emerald 2 which was presented at TCT it was published in Jack Imaging just about a month ago uh and in it he explored the complimentary roles of stenosis plaque morphology burden of plaque as well as hemodynamic forces as delineated by sheer stress quantitatively using computational fluid dynamics and also Delta ffr and interestingly similar to the emerald 1 Study it would appear that the Delta ffr the very measure that I highlighted at the beginning of the talk was important to help identify a lesion that will derive benefit from revas as it relates to angina relief may also be a marker of a lesion that's at higher risk of rupture based on the mechanical forces of the trans trans lesional pressure loss so I think there's certainly some real interest in in complementing mechanical forces with plaque features in predicting Downstream risk but maybe I'll go go to both of you and and feel free either one of you to to speak first but whether it's uh topological changes or features you know we've seen a lot of work in radiomics or perivascular fat or any other any other ideas you have how we can further try to refine our understanding of the uh the vulnerable Legion I I'll start by saying I think it's a very challenging concept I think in cardiology this term vulnerable plaque that was you know coined actually my institution uh over over 20 years ago uh is is kind of been a holy gra Bose of noninvasive and invasive Imaging but the problem is that you need a very good positive predictive value a good specificity to be confident enough to say yes that is really high enough risk that you're going to do something locally a local therapy extent um in the reality as I've I've yet to see data that tells me that so I I personally maybe I'm a little biased as a preventive cardiologist I think of atherosclerosis as a systemic disease and a systemic treatment and if I see a high-risk plaque I say we need to kind of double down in that those preventive therapies uh and treat the patient even more um but I think your question and I think the question from the participant is you know is plaque Imaging is plaque morphology going to be able to make us even better to address that her lesion uh risk and you know perhaps I I think it's going to require ultimately a combination of both Imaging data and non-imaging data I think I would combine it with you know risk factors with other things but um but I think we're at least moving in that direction that we're getting better and better at looking at those high-risk lesions but I'm it's still open to me whether we're going to have a high enough specificity and ultimately it'll be a trial that's needed to take folks that have what we think is a high enough risk in randomiz them to a a focal therapy versus not yeah really thoughtful comments I'll let you bring us home with your thoughts on that Michael but uh uh you know what I always tell people when they ask me this is that you know it's it's not been achieved with intravascular imaging with all the spatial resolution and image Fidelity but I do I am excited about the ability to of course take all these tools and apply them to the question to a non-invasive data set in a unique way right as you mentioned whether it's mechanical forces or plaque or or fat inflammation and Michael did you have anything you'd want to share on this topic sure thanks Jonathan yeah I think I would maybe add on to what you mentioned about you know Emerald 1 Emerald 2 Paradigm study some of what even um uh Ron shared uh about Scott hard and the low attenuation plaque and the Sevenfold increased you know risk prediction for folks with a a plaque score greater than 4% or a cubic plaque volume of more than 238 cubic millimeters and what I've enjoyed is looking at these studies when they look at all these AI enhanced um uh features including uh the the pericoronary adapost tissue um inflammation as you described the PCAT when you look at that total plaque volume aoma percentage um you know low attenuation plaque when you look at those features as calculated by AI when you look at the incremental Improvement above percent stenosis absolute ffrct traditional risk factors that we all grew up with you know when you look at any clinical traditional risk factor and it's quite striking that these AI variables stand out time and time again in these trials that look at you know one and three-year uh uh uh cardiac uh uh events so I think it's very provocative I think it's early but I am uh very excited to see the complimentary role of looking at AI enhanced uh features together with the the standard clinical risk factors as well as the time tested percent stenosis and ffrct so it's an exciting topic excellent well uh let me take this opportunity then to thank you both to thank obviously the SC professionalism of the staff is uh really amazing uh as you may know I was the president of the society in 2015 and to see the maturation of the society and its development has been a source of great pride I had very little to do with that but nonetheless I I derive great pride and and I'm so grateful to be part of this Society I want to thank the team from heartflow for sponsoring not only this webinar But continuing to support the scct it's a obviously a fundamentally important Society to our field and to the patients that we serve uh and finally M let me close with a a big a big uh expression of gratitude and and thanks to my dear friends uh colleagues and in many ways mentors DRS Gallagher and blankstein uh really always a pleasure to to uh learn from you and work with you uh so thank you again and to all of you who joined in uh thank you for joining and for sharing your thoughts uh all the best everyone
2024-09-24 00:12
