Evaluation of Coronary Inflammation with CT angiography and Fat Attenuation Indexing

Dr. Nero discusses evaluation of  coronary artery disease and coronary inflammaiton with Dr. Cheerag Shirodaria from the University of Oxford.  Dr. Shirodaria along with Prof. Charalambos Antonidies at Caristo Diagnostics have developed a ground breaking technology that allows us to evaluate not just the severity of a coronary artery lesion, but also how inflamed it is and how likely it is to cause fatal and non-fatal myocardial infarction.  In addition, since it is a biologically active "marker" the Fat Attenuation Index (FAI) can also be used a a way to evaluate success of medical intervention.  Using computer post processing from standard 64 or 128 slice CT coronary angiography, they have been able to evaluate not just the vulnerable patient but also (possibly) the vulnerable plaque.

Transcript

 Evaluation of Coronary Inflammation with CT angiography and Fat Attenuation Indexing

Dr. Thomas Nero: Good morning. I'm Dr. Thomas Nero and welcome to Future Pulse, today. We're going to be discussing CT coronary angiography and evaluation of inflammation with Dr. Cheerag Shirodaria. Dr. Shirodaria is the CEO and co-founder of Cristo diagnostics. He's also an honorary consultant cardiologist at Oxford University. Dr Shirodaria was a cardiology registrar and interventional cardiologists prior to obtaining his MBA and returning Oxford to start the CT imaging program. He co-founded Cristo diagnostics with Professor Charris Antonidies with whom he was a research fellow in the early 2000s

Their work focused on the role of inflammation in the development of coronary artery disease, and also how diagnostic imaging can help predict the outcome of coronary artery disease.

Welcome Dr. Shirodaria. You developed a novel technique called fat attenuation indexing for the evaluation of coronary inflammation. This is utilized as an add on to CT coronary angiography. Can you walk us through this technology and how it works? 

Dr. Cheerag Shirodaria: We've known for over 50 years, that atherosclerosis is an inflammatory disease. And our problem is that we've never had a straightforward way of detecting and monitoring coronary inflammation. We've known that inflammation drives not only the development of plaque, but it also drives plaque rupture.

And although there've been some important landmarks with regards to blood-based biomarkers, such as high sensitivity, C-reactive protein, which has been used in a number of studies, our problem has been that while they are very good at a population level, , when you try to calculate an individualized patient risk, then it becomes a little more difficult.

The work of my fellow co-founders at Christo and all of us are cardiologists who've been focused, on atherosclerosis and vascular biology research , has really been to highlight the importance of peri-vascular adipose tissue.

 If you look back at the literature, the assumption had always been that the fat tissue that surrounds blood vessels, the so-called peri-vascular adipose tissue, the hypothesis had always been that, , the adipose tissue somehow secretes harmful molecules, and is detrimental to the coronary arteries and may actually provoke and stimulate the development of atherosclerosis, but the academic work undertaken in the scientific lab here at the university of Oxford by my fellow co-founder Prof. Charalamros Antoniades, who is really the brains behind this actually showed that, the reality is actually quite different. What we showed, when you have an inflamed coronary, Chemical signals are released from that inflamed coronary artery so cytokines such as IL-6, TNF-alpha, gamma interferon and these cause changes to the texture and composition of the surrounding peri-vascular adipose tissue. In short, what we found was that adipocytes become smaller, and the ratio of fat to water content changes within them. 

The real leap that was then made was once this disease process had been established, from a basic science perspective, our next challenge was to think about how this could potentially be imaged and captured so that we could actually translate this into something that we could use in our patients.

Now Oxford has a very proud reputation in cardiac MRI, and we initially looked at cardiac MRI with my colleague and fellow co-founder Stefan Neubauer to see whether Cardiac MR could play a role in helping us visualize these changes in adipose tissue composition and texture, but quite simply, the temporal and spatial resolution of Cardiac MR wasn't good enough. So we then turned our attention , to coronary CT angiography. Around that time the guidelines in the UK had changed, to really focus on using CT as a first-line imaging test for patients with suspected coronary artery disease, and what we found was actually that CT was a perfect imaging modality because it had the required resolution to not only visualize the coronary arteries, but also to allow us to visualize these changes in the texture and composition of the perivascular adipose tissue

 We published our findings in Science Translational Medicine, in 2017. And there was a lot of interest, and scientific fanfare over the results because it's really was a very novel way, of imaging coronary inflammation.

But there was some very important questions that the cardiology community and the broader scientific community asked of us, where the findings that we found both real and reproducible and we were able to show that indeed, this was.  

And the third question, which was asked in a very elegant, a commentary piece, in Nature Cardiology by Zahi Fayad and colleagues. So, what and the, so what was really, this is a very novel and an elegant way of detecting coronary inflammation, but does this help us predict who's going to have a heart attack and who's not.  That was published the following year , in the Lancet where we published the CRISP-CT trial. , CRISP-CT , was a 4,000 patient , post-hoc analysis , of two independent, prospective registries, one from Europe, one from the U S and , these were unselected patients undergoing, clinically indicated coronary CT angiography and what CRISP-CT showed was that patients who had increased coronary inflammation, , as evaluated using our biomarker, which is called the fat attenuation index or FAI, those patients had a six to nine fold, increased risk of having a fatal heart attack and a fivefold, increased risk of having a non-fatal heart attack. And this was after adjusting for all conventional risk factors, such as age, diabetes, cholesterol, but also imaging characteristics, such as plaque burden and calcium scoring.

Tom: And that was just using a cutoff of 70 Hounsfield units, but you also have a linear regression analysis that showed an even closer association. 

Cheerag: It's correct. So what we did, in the training and validation of the methodology, we determined the optimal cutoff point, both from a sensitivity and specificity perspective. And that turned out to be a Hounsfield unit level of minus seventy. Now the story isn't quite that straightforward because each individual patient is their own control.  And nature is not binary. Inflammation is a continuous variable, but fundamentally what we showed with the patients who had increased coronary inflammation were at a significantly increased risk of developing both fatal and nonfatal, myocardial infarction over the following 10 years over and above all other risk factors and importantly in a post hoc analysis that we published in JACC the following year, we showed that when we analyzed coronary CT angiograms we look for so-called high risk plaque features, which we look for in a subjective manner.

Things such as spotty, calcification, low attenuation plaque, napkin ring sign. And what we showed was that FAI was actually much more predictive of outcome than the presence of high-risk plaque features. And what was very important was that in patients who did not have high-risk plaque features, but did have increased coronary inflammation, they did very badly.

It's telling us something about the activity of plaque in terms of how detrimental it may be.

Tom: And with this, am I correct in saying that you're both identifying the vulnerable patient, but potentially also identifying the vulnerable plaque? 

Cheerag: It's always been the holy grail of cardiovascular disease to identify as you rightly put it, both the vulnerable patients, but ultimately the vulnerable plaque. So what we've done with the device so far, and this has been turned into a regulatory clear device, which is CE marked in Europe, also regulatory cleared in Australia and will surely be approved by FDA.

What we've shown is that we can identify the vulnerable patient. Now, the next question, which is a very pertinent one that you ask is: Can you evaluate individual plaques? In other words, can you say that proximal LAD lesion that may only look like a 50% lesion now is actually highly inflamed and we'll go on to cause a subsequent event in that patient that's work. That's currently ongoing in the academic setting in Oxford, which Charis Antoniades is currently leading. And we're working with other groups around the world looking at this, but we now have the ability to actually interrogate individual plaques to determine how inflamed they are to answer that exact question.

And that's what we hope to do over the next year or so.

Tom: I'm sure there's going to be some criticism saying that while even if you do identify that vulnerable plaque that doesn't necessarily mean that we need to go into intervene. And I wholeheartedly support that even though I am an interventional cardiologist, but that the answer is not. stenting, no, the answer is actually treating the underlying disease process and we need to really be focusing on that.

And, if we can do that, then we can really put a lot of the procedures that we do in the past and really go forward and treating the patient. One question I had as I was thinking through your technology, which I think is absolutely fascinating, is that what you've shown here is that you can identify the effects of the inflammation.

Has any of your research been able to identify the cause of the inflammation or what's bringing that on? And as a secondary question, What will that then mean towards our pharmacologic therapy? That the pharmacologic therapy really is about anti-inflammatory therapy or are there other mechanisms that we should also be looking at? 

Cheerag: So one of the reasons why coronary artery disease remains the number one cause of mortality and morbidity in the world is because it's multifactorial. Now, what we do know, is that the ultimate end product of various disease processes being diabetes, be it hypercholesterolemia, be it genetic predispositions.  They all ultimately cause inflammation. So what we focus on is measuring, the end product of what is happening rather than necessarily the multiple causes. Because as I said, they all end up causing inflammation. What's that allows us to do is to identify risk associated with that. And, that's a very important point, but I think , the key question is it's important that we've shown that what we can measure is both real and reproducible , and is predictive of outcome.

But when we have conversations with our patients, it's not really very helpful to tell them what their risk of dying is, or the risk of having a heart attack is unless you can do something about it. And as you rightly point. The answer is not necessarily stents. It's going to be a combination of other things.

Now, one of the big problems that we have in cardiology is that when we actually start our patients on treatments, we actually have very little idea. Whether those treatments are actually modifying our individual patients risk. We can measure two things that can give us objective evidence of that.

And that's LDL cholesterol and the lowering and systolic blood pressure, if you look beyond that, it's quite difficult for us to determine whether individual treatments, are having an effect on risk. Now, our biomarkers are dynamic. They actually change in response to drug treatments, in our Lancet paper, when you looked at the thousand or so patients who had increased coronary inflammation who started statins and, or, aspirin and you've compared them with a thousand or so patients who did not start stat into aspirin, but had increased inflammation.

Their outcomes were very different. So, for the patients, who've had increased inflammation who did not start treatments, their hazard ratio was 18. In other words, an 18 fold increased risk of having a fatal MI whereas those who did start treatments. They're hazard ratio fell to just below three and actually lost statistical significance.

So that's the first bit of evidence showing that treating FAI and treating increased inflammation modifies risk. The second bit of information is there's been studies published showing that statin treatment modifies FAI. So giving statin treatment for eight to 10 months with baseline and follow-up CTs, we can see that inflammation reduces, which is hugely important.

The third, bit of evidence that we have is that we actually conducted a study with the NIH who have an interest, in rheumatological and connective tissue diseases, and associated cardiovascular risk in particular psoriasis. So we undertook some work with them they compared biological therapies with non-biological treatments, such as methotrexate.

And what they showed was that treatment such as methotrexate had no effect on coronary inflammation. Whereas biological treatments have just TNF alpha had significant effects, on modifying and reducing inflammation. I think the other thing that's very important, is for the majority of patients, simple treatments, but identifying the right patients who we should be treating with statin and aspirin, I think is the important thing.

Because as much as we all recommend these things to our patients, our patients don't always want to take treatments and showing them objective evidence that they are at increased risk because of increased inflammation and that risk is modifiable is very powerful for us as cardiologists, both to monitor, but also to share that information with our patients.

Tom: It seems to really sit in that sweet spot of Bayesian analysis that if you do a CTA and they have no plaque, that they're fine. And if you see someone who has a lot of coronary calcium and a significant diffuse plaque, you are going to be treating them aggressively, but for that middle third of patients for middle 50% of patients who do have modifiable risk factors, that these are the ones that you're going to be focusing on.

Cheerag: It's been very interesting as this technology is, as I said, has been used in Europe. And we've been looking at data now in hundreds of patients who've been having the test and it's been quite eye opening if we were to take an unselected patient population, having coronary CT angiography who had minimal coronary artery disease.

We're actually finding that about 17 to 25% of those patients have increased coronary inflammation. And that's defined as an FAI score, which is a corrected value of FAI because you have to correct for technical and patient-specific factors between different CT machines, between different patients.  If their inflammation level is above the 70th percentile, there are at least two and a half times more likely to have a fatal MI.

So just to reiterate, we're finding that about 17 to 25% of patients who have essentially pretty normal coronary arteries, who certainly me as a cardiologist, I'm not necessarily recommending aggressive treatment to those patients. We're actually finding that increases their cardiovascular risk by at least two and a half times.

Tom: You also have some other, novel, technologies that you're developing. One is PVAT P V A T, which is the radio autonomic characteristic of the peri vascular adipose tissue. And there's some theory that this will help you to be able to determine acute plaque rupture from chronic plaque, from acute inflammation to chronic inflammation.

How will this fit into the future of what you are developing? 

Cheerag: So, one of the things that's now possible because of advanced computational power, we can now extract much more information from an individual image. So from each voxel or pixel of a CT scan, we can extract about 2000 different bits of information.

But what we've been able to do at Cristo and in particular at the university of Oxford is that we have a unit by resource in Oxford where we have thousands of patients. Who've undergone cardiac surgery where we've collected samples of myocardium, vascular tissue, adipose tissue from 10 to 15 different depots around the body.

We've extracted the RNA. So these patients have been characterized genotypically phenotypically, metabolically proteomically, radiomically with outcomes, data, allowing us to build what we call radio transcriptomic signatures. This is a new phrase that my colleague Charris kind of coined.

What we're trying to really combine is deep phenotyping of the patient from a basic science perspective, along with deep phenotyping, from an imaging perspective , to develop novel biomarkers and measures of risk and inflammation. So that's allowed us to really tease out new insights into the difference between acute and chronic inflammation.

We know histologically that when you have chronic inflammation, in the adipose tissue, that leads to not only Neo angiogenesis, but it also leads to deposition of collagen, fibrosis, and we've been able to detect that not only in cell culture, looking directly at the adipose tissue, but we've also been able to construct signatures of this as to what it looks like on a CT image.

We can distinguish between acute and chronic inflammation.  We've also published some data in COVID 19. We know COVID-19 is an acute inflammatory disease, ultimately, and we constructed a custom-made imaging bio-signature, of COVID-19 associated acute inflammation, which was highly predictive of outcome.

 This was presented at the AHA  about 18 months ago and will shortly be published. And we can do that with many different disease processes. With a drug, we can custom make a signature, an imaging signature of how that drug and its biological processes will affect, and change imaging characteristics.

Tom: That's absolutely fascinating. The future of what you have with this technology. it's hard to imagine all the different possibilities of where this could go. You had mentioned that there is association between the inflammation that you're seeing and serum biomarkers.

Why don't we just measure cardiac CRP? How much better is this than CRP and what are the downfalls of just relying upon systemic inflammatory biomarkers?

Cheerag: I'd love to say that the best way for us to measure coronary inflammation is with a systemic biomarker. It'd be a hell of a lot easier for us in clinical if that was possible. I think the problem is that if you actually want to know, what's going on in the actual coronary arteries, systemic biomarkers, such as CRP are simply not sensitive or specific enough. And we've actually shown that the correlation between our measures of coronary inflammation and CRP are actually very weak at around the 0.1 level. Fundamentally I think we're measuring something different than what systemic biomarkers are able to capture.

Tom: Yeah, I've always been intrigued that vascular disease is a local process. You can have a patient who has severe peripheral arterial disease and have absolutely no coronary artery disease. And it just doesn’t completely makes sense , except that there must be something more locally happening. That we don't completely understand exactly what the process is. I think this is going to bring us a long way in getting to that diagnosis and then trying to work our way backwards into what the initial causes of these issues are.

On some technical points. You can do this on both 64 slice and 128 slice CTs? 

Cheerag: So one of the really, really cool things about this technology is that we can evaluate coronary inflammation on any cardiac CT or a scan acquired on at least a 64 slice scanner. So indeed, our original data, we actually analyzed in scans acquired in 2005. So very early 64 slice scanner technology.

So as long as the scan meets SCCT requirements, we could do the analysis. We don't require. technologists cardiologists or radiologists to alter any of their protocols. All of our analysis has made on the standard image and indeed we validated it very extensively across different CT manufacturers, different platforms, different reconstruction, algorithms, different technical factors.

And that's really why the crude measure of FAI. This simple binary minus 70 cut-off isn't is not helpful. Clinically, there are a number of things that we have to correct for, which is why we produce something called the FAI score presented on centiles the corrects rule, these factors, and that's taken a lot of development, , to ensure that we can correct for all those factors.

Tom: Is heavy calcification a problem for you. When we look at other CT imaging protocols and post processing like, FFR-CT, heavy calcification does pose a problem. We've always talked about the, fact that calcium appears to be a risk factor,

it's not the calcium, it's the plaque. It's not the plaque, it's the inflammation in the plaque. But unfortunately with FFR CT, you can't get good imaging quality if you have heavy calcification, does your technology overcome that?

Cheerag: The simple answer is yes. And I'll explain why. So as you rightly point out plaque imaging and stenosis imaging is heavily constrained by both calcification and also the presence of stents. And the key thing, with our technology is because we're looking at the perivascular space.

So we're actually looking outside of the coronary arteries, our technology is not affected by other calcification or the presence of stents. And indeed we and published on that in the European heart journal.

Tom: Are there any limitations other than good quality CT scan or other times when you say, you know, this isn't going to be the right technology for this. 

Cheerag: It's a really important question.  In real world settings our rejection rate is about one to 2% and it's primarily due to motion artifact.

Tom:  One of the questions has been coming up a lot is our ability to look at diversity within our patient populations and especially in the research setting to ensure that we have adequate diversity. And one of the discussions that we've been having is are we really addressing the question of diversity, as far as what our risk predictions are and then the calculations for outcomes. So for example, if someone is put into a research study and has a lower than actual predicted risk, their outcomes would be magnified. Have you looked at the way that your technology can inform our ability to use predictive modeling? 

Cheerag: so, again, as someone who spent most of the last decade working in drug development,  the issue of both racial and gender diversity in trials is an extremely important one. And probably one that we haven't paid enough attention to. From our angle, at Caristo and from a scientific angle in Oxford, this has been a very important area of focus.

So, the university of Oxford is actually building and has built the largest CT registry in the world. So the ORFAN registry aims to have 250,000 patients from across the world with outcomes, data, and, we're well over 100,000 patients in the registry already, and that is allowing us to address those important questions around ethnicity, from what we've seen so far in our preliminary data, it does not seem to be a significant factor.

However, We will delve into this very important question in a lot more detail, over the next 12 to 18 months, 

Tom: I think it's going to be extremely important to find out, and it's going to be really helpful, not just in the assessment of cardiovascular disease, but I do think it's going to be extremely helpful in the way that we look at and treat patients overall.  What you've just brought up is, one of the big questions about social determinants of health and that this will go a long way to understanding these issues.

 So, two final questions for you:  First, what are the next things that you guys are looking into? What's the future of Christo and your and Charris' research? And where do you think that this is going to lead us?

Cheerag: The focus I think, is on a number of fronts. Firstly, trying to understand better, the relationship between. Inflamed plaque or non-inflamed plaque and identification of that. And what that really means in clinical practice, that's one definite area of focus. The other two areas of focus are the biological and scientific discoveries that we've made in Oxford over the past decade.

In other words, the importance of adipose tissue inflammation has great relevance in other disease areas apart coronary artery disease. And so the other two areas that we're focusing on at the moment and indeed are developing products is in atrial fibrillation and stroke prediction.

Now your listeners will be well aware that the commonest cause of ischemic stroke, is actually atrial fibrillation. And we know that atrial myopathy is an inflammatory disease. And so we have developed technology, that looks at pari-atrial inflammation, and that's is predictive of both AF and future stroke risk.

So that's currently being validated in a very large series of studies funded by both the British heart foundation and also a very large horizon, 2020 consortium in Europe called the Maestria consortium. That's really looking at novel markers of AF and stroke risk. And so that we hope to bring to market in the next 12 to 18 months.

The second area is looking at subcutaneous,  epicardial,  Thoracic fat depots because we know that adipose tissue inflammation is very important, but doing fat biopsies on people is A not practical and B you can't actually access fat depots, but actually we know that the degree of inflammation in these fat depots is highly predictive both of the development of diabetes and outcomes. So, we developed technology essentially to do a virtual fat biopsy, and that's been developed in collaboration with the UK government, through an NHSI award. That's been validated in about 30,000 patients. So we hope to bring that to clinicians,to help make better decisions about identifying risk, associated with adipose tissue inflammation very shortly

Tom: And then the final question, which I'd love to ask everyone is. Your work is brilliant and exciting, and has really gotten me excited whose work do you look at that you find that's exciting and who are you looking at to say, yeah, you know what, I'm going to follow this guy because what he's producing or what she's producing is really interesting and I would love to see more of what's happening there. 

Cheerag: The thing that really interests me at the moment is that we need to become much better at prevention, keeping people out of hospital and managing risk better and earlier. Now I think that the technology that we've developed here at Cristo is very important, when you have a CT image.  When you have an image that you can evaluate and quantify inflammation in, but I'm very interested in what's going to come further upstream in terms of identifying risk, at a patient level. And I think things like polygenic risk scores, I think could have a really important role to play at a population level, should everyone at the age of 25 or 30 be getting a blood test to look at risk for a variety of diseases.

From a coronary artery disease perspective, we hope that leads to an assessment of inflammation and what's actually going on in the coronaries. I'm really fascinated to see how we merge the two, because fundamentally this will profoundly change the way that we manage disease over a 20 to 30 year period, it requires clearly the evidence to be there, but it will require health systems to be very bold and actually having the balls to focus on prevention rather than only when patients actually have established disease, which is unfortunately what we spend most of our time doing.

Tom: In addition, there's going to be a lot of concern about the privacy rights and how we're going to be dealing with this information because the breadth of the information that we're going to be getting from this is mind boggling. It's it is staggering. And I recently had a conversation with a healthcare lawyer at Wake Forest University and she's working on the, the legal basis of how we protect people's information both in the research setting and in the clinical setting as well. 

Cheerag, I can't Thank you enough for spending all this time and doing this for us today. I can't wait to see what happens in the next year. And I look forward to meeting up with you and Charris, hopefully at the European society meetings in August , and lifting a glass to toast both of you for all your great work. Thank you.

Cheerag: Thank you very much.