Circulation August 24, 2021 Issue

This week's episode features special Guest Host Victoria Delgado, as she interviews author Qiang Zhang and Greg (who was the editorialist and handling editor) as they discuss the article "Towards Replacing Late Gadolinium Enhancement with Artificial Intelligence Virtual Native Enhancement for Gadolinium-Free Cardiovascular Magnetic Resonance Tissue Characterization in Hypertrophic Cardiomyopathy."

Dr. Carolyn Lam:

Welcome to Circulation on the Run, your weekly podcast summary and backstage pass to the Journal and its editors. We're your cohost. I'm Dr. Carolyn Lam, associate editor from the National Heart Center and Duke National University of Singapore.

Dr. Greg Hundley:

And I'm Dr. Greg Hundley, director of the Pauley Heart Center at VCU Health in Richmond, Virginia. And Carolyn, this week's feature actually is a magnetic resonance imaging paper, which you know, of course, I'm very fond of. And these authors have come up with a new methodology to get information equivalent to a late gadolinium-enhanced exam without administering gadolinium. But before we get to that feature discussion, how about we start in with some of the other articles in this issue?

Dr. Carolyn Lam:

Yep.

Dr. Greg Hundley:

Well, how about if I go first? So Carolyn, the first paper I'm going to discuss is from Dr. Paul Welsh from the University of Glasgow, and it pertains to abdominal aortic aneurysms, which you know can occur in patients who are ineligible for routine ultrasound screening. A simple abdominal aortic aneurysm risk score was derived and compared to current guidelines used for ultrasound screening of abdominal aortic aneurysms.

Dr. Greg Hundley:

And so, this study comes to us from the UK Biobank, and they examined participants without previous, and let's just abbreviate this as AAA or AAA. So without previous AAA, we're split into a derivation cohort of 401,820 individuals, and a validation cohort of 83,000 individuals. An incident AAA was defined as a first hospital, inpatient diagnosis of AAA, death from AAA, or a AAA-related surgical procedure. And, of course, they used multivariable Cox models to develop the derivation cohort, and then apply that to the validation cohort.

Dr. Carolyn Lam:

Wow, Greg, that is a large number of people, the power of the UK Biobank, huh? So, what did they find?

Dr. Greg Hundley:

Right. So, Carolyn, components of the AAA risk score were age, stratified by smoking status, weight, stratified by smoking status, any hypertensive and cholesterol-lowering medication use, height, diastolic blood pressure, baseline cardiovascular disease, and then diabetes.

Dr. Greg Hundley:

So, Carolyn, in the validation cohort, over 10 years of follow-up, the C index, for the model for the USPSTF guidelines, was 0.705, whereas, the C index of the risk score as a continuous variable was 0.856. And in the validation cohort, the USPSTF model yielded a sensitivity of 63%, and a specificity of 71%.

Dr. Carolyn Lam:

Okay, Greg, but what's a take-home message?

Dr. Greg Hundley:

Right, Carolyn. So the take-home message is that in an asymptomatic general population, a risk score based on patient age, height, weight, and a medical history may improve identification of asymptomatic patients at risk for clinical events from AAA. And also, these results highlight that further development and validation of risk scores to detect asymptomatic AAA are needed.

Dr. Carolyn Lam:

Wow, and that was a great summary with a lot more data in the article, huh? Let's refer the readers to it. But for my paper, it highlights the key role of histidine triad nucleotide-binding protein 1, or HINT 1, in the pathogenesis of cardiac hypertrophy.

Dr. Greg Hundley:

Wow. Carolyn, so tell me a little bit more about HINT 1.

Dr. Carolyn Lam:

I thought you may ask. HINT 1 is a highly-conserved 14 kilodalton protein that belongs to the histidine triad super family. It was previously shown to play a role in diverse neuropsychiatric diseases. Loss of HINT 1 increased susceptibility to carcinogenesis in mice, suggesting, as well, a tumor suppressor role. So recently, HINT 1 has emerged as a tumor suppressor with multiple molecular mechanisms, involving regulation of apoptosis, gene transcription, and cell cycle control.

Dr. Carolyn Lam:

Now, with that as a background, today's paper from co-corresponding authors Drs. Ji, Xie and Han, from Nanjing Medical University, used animal models and cell models of hypertrophic growth, and found that HINT 1 deficiency aggravated overload-induced cardiac hypertrophy and fibrosis and deteriorated cardiac dysfunction in mice, whereas, cardiac-specific HINT 1 over-expression attenuated cardiac hypertrophy, and rescued cardiac dysfunction.

Dr. Carolyn Lam:

Furthermore, the more the authors uncovered Homeobox A5, or Hox A5, as a HINT 1 target gene, which contributed to hypertrophy through activating the TGF beta signal pathway. Combined, these findings demonstrate HINT 1 may be a prognostic biomarker, and this may establish a foundation for future investigation of its potential as a therapeutic target for cardiac hypertrophy and heart failure.

Greg Hundley:

Great, Carolyn. So, we get a hint for future cardiac hypertrophy.

Dr. Carolyn Lam:

Hahaha.

Dr. Greg Hundley:

Couldn't resist.

Dr. Carolyn Lam:

Bravo.

Dr. Greg Hundley:

Couldn't resist. So, Carolyn, my next paper comes to us from Dr. Mark Gladwin from the University of Pittsburgh. And, as you know, many patients with one of your faves, heart failure with preserved ejection fraction have metabolic syndrome and develop exercise-induced pulmonary hypertension. Now, increases in pulmonary vascular resistance, in patients with HFpEF portend a poor prognosis. This phenotype is referred to as combined pre- and post-capillary pulmonary hypertension.

Dr. Greg Hundley:

And therapeutic trials of exercise-induced pulmonary hypertension, and pre- and post-capillary pulmonary hypertension, have been disappointing, suggesting the need for strategies that target upstream mechanisms of the disease. And so, this work reports novel rat exercise-induced pulmonary hypertension models and mechanisms of pulmonary vascular dysfunction, centered around the transcriptional repression of the soluble guanylate cyclase enzyme in pulmonary artery smooth muscle cells.

Dr. Carolyn Lam:

Ooh, so much of this is of interest to me, from HFpEF to soluble guanylate cyclase. Ah, all right, so what did they find, Greg?

Dr. Greg Hundley:

Right, Carolyn. So, in the HFpEF and in the pre- and post-capillary pulmonary hypertension models, metabolic syndrome contributed to pulmonary vascular dysfunction and exercise-induced pulmonary hypertension through enhanced reactive oxygen species and MIR193B expression, which downregulates NFYA-dependent soluble guanylate cyclase beta-1 expression, and adenovirus mediated NFYA over-expression and SGLT-2 inhibition restored NFYA soluble guanylate cyclase beta-1 cyclic GMP signaling, and ameliorated exercise-induced pulmonary hypertension.

Dr. Greg Hundley:

So Carolyn, after all of that, these results uncover a molecular explanation for the unsolved clinical associations linking metabolic syndrome with exercise-induced pulmonary hypertension in patients with pre- and post-capillary pulmonary hypertension, as well as HFpEF.

Dr. Carolyn Lam:

Wow. That is super, Greg. Thanks. Now, other articles in today's issue include a research letter by Dr. Schunkert on identification of a functional PDE5A variant at the chromosome 4q27 coronary artery disease locus in an extended myocardial infarction family.

Dr. Carolyn Lam:

There is a policy front paper by Dr. Lackey on applying decision analysis to inform FDA's benefit risk assessment of ticagrelor for primary prevention of myocardial infarction or stroke, based on the THEMIS trial. There's an exchange of letters between Drs. Olson and Reiffel regarding the article emulating randomized controlled trials with non-randomized real-world evidence studies, the first results of the RCT Duplicate Initiative.

Dr. Greg Hundley:

Great. Carolyn. Well, I've got two other papers. There's an ECG challenge from Professor Macherey entitled, Wide QRS complex bradycardia in a hemodynamically unstable young woman. And then, finally, one of our nice perspective pieces from Dr. Armstrong entitled, Extending the product label for ticagrelor: Looking under the FDA hood. Well, Carolyn, now we're going to learn a little bit more about that non-contrast method of MRI to get, maybe, the equivalent to late gadolinium enhancement in patients with hypertrophic cardiomyopathy.

Dr. Carolyn Lam:

And all by AI. Very, very cool. All right, let's go.

Dr. Victoria Delgado:

Hello, I'm Victoria Delgado, associate editor of Circulation, and working at the University Medical Center, and I have the privilege to welcome Dr. Zhang from the University of Oxford, division of cardiovascular medicine, and first author of the article published in Circulation, which presents a new technology that has the potential to change our clinical practice, and how do we assess myocardial tissue characteristics, particularly fibrosis with cardiac magnetic resonance.

Dr. Victoria Delgado:

With us, we have also Dr. Greg Hundley from VCU Pauley Heart Center, associate editor of Circulation, and host of Circulation on the Run and editorialist of this article. And Greg has handled this article in the review process and guided Dr. Zhang and coworkers to finally get a very impactful, in my opinion, and novel paper that has the potential to change clinical practice. He will help us to put in perspective the main message of this article. But first, Dr. Zhang, why don't you tell us briefly, what is this new technology for myocardial tissue characterization with cardiovascular magnetic resonance, and why did you develop it?

Dr. Qiang Zhang:

Well, first, I want to say thank you very much for the invitation. Well, as we know, CMR late-gadolinium enhancement, or LGE, has been the imaging gold standard for micro tissue catheterization. However, LGE requires the injection of a contrast agent, which prolongs the scan, increase the cost, and is cautioned in some patient groups. It is therefore desirable to develop a contrast-free technique to replace LGE. A native or pre-contrast CMR such as cine imaging, T1-mapping and the T2-mapping are alternative means for myocardial tissue characterization without the need for contrast.

Dr. Qiang Zhang:

And notably, T1-mapping can detect a wider range of pathology, but its clinical application is hindered by confounding factors and a lack of clear interpretation. So we thought that since MRI is inherently multimodal and multiparametric, with different modalities reflecting complimentary information about micro tissue, therefore can we use the ones of an AI  method to combine the enhanced pre-contrast and modalities to produce a virtual LGE image without the need for contrast, and this leads to our concept of CMR virtual native enhancement.

Dr. Victoria Delgado:

Excellent. And in which population did you evaluate this technology? Why did you choose? So, can you tell us a little bit the characteristics of the population where you choose to assess the performance of this new technology?

Dr. Qiang Zhang:

Yes, so we validated this concept first on the hypertrophic cardiomyopathy patients using the image data from HCMR study, which is a larger multinational study. We trained the new network models on about 2,700 images, and then we tested independently on 124 patient materials.

Dr. Victoria Delgado:

And what were the results? Can you summarize them for us to understand what you found, how you evaluated the performance of this new technology compared to classical late gadolinium enhancement?

Dr. Qiang Zhang:

Yeah. So, we found that first, the VNE image had significantly better image quality than the traditional LGE, and the secondly, the VNE revealed characteristic HCM lesions in hypertrophic segment and added the anterior and the inferior right ventricular insertion point, and those VNE lesions were in high visual spatial agreement with the lesions detected by LGE. And thirdly, VNE correlated strongly with LGE in quantifying hyperintensity, micro lesions, but also more subtle intermediate-intensity lesions as often observed in the HCM patients.

Dr. Victoria Delgado:

Exciting. And now I would like to turn to Greg. You are a renowned expert on CMR. What did this article attract you? What were the main findings that you found interesting in this article?

Dr. Greg Hundley:

Yes. Thank you so much, Victoria. And just first, before I get started, I wanted to thank Zhang for sending us this work, and his entire team really spanning several institutions around the globe. Victoria, I want to thank you for leading this discussion. And Victoria, we work as a team looking at imaging, and what's very impressive is this caught multiple of us on the imaging team, but also the editorial board. So want to thank you Victoria. And then, finally, I want to thank Dr. Charlotte Manisty and Jennifer Jordan who helped with the editorial.

Dr. Greg Hundley:

So, just to take a little bit of a step back in the year 2000, Dr. Ray Kim, Dr. Bob Bono, Dr. Bob Judd at Northwestern University, in Chicago, administered gadolinium contrast. It's an extra cellular agent and they were examining myocardial perfusion, and then they noticed if they took images 10 or so minutes after that administration, they appreciated this late enhancement of the myocardial tissue that was associated with two things, one myocardial injury, and then, also, scar formation.

Dr. Greg Hundley:

And for the last 21 years, that technique has been very valuable in assessing infarct size in those with ischemic cardiomyopathy, and then also recognizing extracellular fibrosis or forms of myocardial injury in a variety of non-ischemic cardiomyopathy processes. And one of those is hypertrophic cardiomyopathy. And so, Zhang and his group have been working on a technique that did not utilize gadolinium anymore. And they based this on, really, had two important features.

Dr. Greg Hundley:

One is, some examination of the T1 relaxation that's available when you acquire magnetic resonance images, and then also applying artificial intelligence to analysis of these images. Those two factors allowed this investigative team to produce images that are very similar to what we appreciate with gadolinium enhancement in patients with hypertrophic cardiomyopathy. And why is that significant? So administration of gadolinium, we think about two things, again, that are an issue with that.

Dr. Greg Hundley:

Well, one, we've got to give the contrast agent, and some patients are not well-suited. If you have renal dysfunction, you can develop nephrogenic systemic fibrosis, a scleroderma-like syndrome when the gadolinium is not cleared, and it's stays in your body a long time. The other thing we tend to worry about much more recently was accumulation of gadolinium in the brain stem. We don't know that that has an adverse effect, but we're certainly aware of it.

Dr. Greg Hundley:

And finally, I guess there is a third point. There are a few patients, one in 20 to 40,000 that have allergic reactions. So, for all those patients that really can't get gadolinium, this is another potential opportunity moving forward. The second point is time. So during an MRI scan time, or process, we administer the gadolinium. And remember what I said, we've got to wait about 10 minutes to then collect these images. Well, if you can have a technique where you don't have to administer gadolinium, you get the same administration and you don't have to wait that 10 minutes, we might be able to save a large amount of time. And you say, well, 10 minutes, what does that mean?

Dr. Greg Hundley:

But if you're doing many patients during the day, that really could equate to a big time-savings. So what attracted us to this article? A technical innovation that perhaps may obviate the need for the administration of gadolinium in many cases, and here in these patients with hypertrophic cardiomyopathy, I think the images were quite stunning in that they appreciated the extent of the fibrosis and scar in patients with hypertrophic cardiomyopathy to the same degree, and maybe even with higher image quality than we had in the gadolinium-enhanced comparitors.

Dr. Victoria Delgado:

Yeah. I agree with all those comments, indeed. For these 20 years, there have been other developments, in terms of assessing diffuse fibrosis with T1-mapping techniques after administration of gadolinium, but also before administration of gadolinium. And this technology Dr. Zhang and co-workers have shown actually reduces the time of acquisition because you only need 15 minutes to acquire the cine images and the T1 images. And the analysis is not very long in terms of post-processing, because it's also very short. As seen the paper was indicated.

Dr. Victoria Delgado:

And as you said, the administration of gadolinium enhancement is not free of potential risks. And in these patients, for example, we have hypertrophic cardiomyopathy that can undergo repetitive evaluations during follow-up. The use of this imaging technique can help to reduce the exposure to gadolinium enhancement.

Dr. Victoria Delgado:

Now, the question for you, Qiang, will be, what are the next steps for you and your team to further develop this technique, to further convince the community to implement this technique in clinical practice, and that can have an impact in our management of these patients with hypertrophic cardiomyopathy, or with other patients on whom we also evaluate the presence of myocardial fibrosis?

Dr. Qiang Zhang:

Thank you, Victoria. And thank you Greg, for the very insightful comments. So we think that the immediate future work would be to validate the VNE HCMR study clinical outcomes as a potential new contrast-free biomarker for HCM patients. And in the meantime, we are working on extending the method to other pathologies, particularly myocardial infarction, for viability assessment.

Dr. Qiang Zhang:

And we also think that the concept of AI virtual contrast agent maybe apply it to other post-contrast images, such as early gadolinium enhancement and extracellular volume refraction mapping, or even, maybe, first perfusion. And also in collaboration with MR vendors, we plan to implement VNE as inline sequence on the scanner to display lesions immediately after the pre-contrast cine and the T1-mapping acquisitions.

Dr. Victoria Delgado:

Very interesting. And Greg, another group of patients, for example, that I'm thinking of where this technology can be very helpful, and you have expertise on this, are patients undergoing treatment with chemotherapy, what we need to address, for example, or to evaluate the presence of cardiac toxicity. Would you see in this population the value of this technique, or how do you see the future of these technique in clinical practice?

Dr. Greg Hundley:

Yeah. Great questions, Victoria. I think probably five things and many of those Qiang has already just brought up. One, with the current study, it's going to be great to follow these patients over time and look at the outcomes. What did amount, presence, location of the findings with this new technique, how did they equate to outcomes in patients with hypertrophic cardiomyopathy? So we're going to anxiously await that.

Dr. Greg Hundley:

I think once we move out of the current study, thinking about other vendors, using this and acquiring images from General Electric and Phillips, and multiple vendors around the world and multiple field strengths, how do we going to understand the findings there? I think another particular issue will be different diseases as Qiang ha said, ischemic cardiomyopathy. Well, what about amyloid? Victoria, you've mentioned using the other possibility here is, can this technique help us produce something similar to extracellular volume fraction measures? And could that be used to identify interstitial disease processes?

Dr. Greg Hundley:

And, for example, as you mentioned, the fibrosis that's associated with the administration of certain chemotherapeutic agents or radiation therapy. I think another thing is, Qiang, can we go back and use retrospectively-collected data? Can your technique be modified so that many of the studies that have been performed in the past, large population studies, like Mesa or a Framingham or Jackson heart study, and could we use that to develop forecasting and algorithms moving forward?

Dr. Greg Hundley:

And then, lastly, I think as you mentioned, the artificial intelligence component of what you've described and how's that can compare when Victoria and I look at the images, and could we get into combining reads from one institution, reads from another, reads from another, tracking outcomes? And so, when a patient comes in and has a study performed, your artificial intelligence, not only does it read that study and highlights the increased signal in the myocardium, but then goes and looks at outcomes across multiple sites, and what would that mean for a given patient with a different condition? So, oh my goodness, the horizon is just so expansive here. The future is very bright. And just to congratulate you on, I really think this could be another landmark technical innovation, so fantastic work.

Dr. Victoria Delgado:

Indeed. Thank you very much, Greg, for all these insightful comments and these perspectives that we have in the future, particularly with the use of this technique in retrospective studies in large cities, that where we can develop these algorithms. And I would like also to thank Qiang for submitting your article to Circulation, for giving us the opportunity to present this new technology and make this article, like the landmark article, where there will be many other articles to follow. We hope that you also choose us.

Dr. Victoria Delgado:

But with this, I would like to thank both of you, Greg and Qiang, for the excellent discussion that we have had in this new technology, based on artificial intelligence, that can identify myocardial fibrosis without the use of gadolinium enhancement, that maybe by now will be obsolete in the future, and that can impact on how we do our clinical practice. Many thanks to all of you and happy to discuss in the future other articles.

Dr. Greg Hundley:

Well, on behalf of Carolyn and myself, I want to thank our speakers, and then also, wish everyone a great week. And we will catch you next week On the Run. This program is copyright of the American Heart Association, 2021. The opinions expressed by speakers in this podcast are their own and not necessarily those of the editors or of the American Heart Association. For more, visit ahajournals.org.