Unlocking Prognostic Information from Cardiac CT: Does Aortic Mitral Continuity Calcification Matter?

See also the article by Willemink et al in this issue.

Richard A. Brown, MBBS, MD, MRCP, is a current fellow in advanced cardiovascular imaging in the department of radiology at St Paul’s Hospital, Vancouver, affiliated with the University of British Columbia department of radiology. He completed postgraduate medical training in the United Kingdom (Yorkshire and London), with higher specialist training in cardiology including a doctorate in coronary artery disease. He has authored a number of peer-reviewed articles and three book chapters related to the pathogenesis and treatment of coronary disease. He is a former associate Royal College of Physicians tutor and current associate editor of European Heart Journal Case Reports.

Jonathon A. Leipsic, MD, FRCPC, FSCCT, is the chairman of the department of radiology for Providence Health Care, Vancouver Coastal Health and the vice chairman of research for the UBC department of radiology. He is a professor of radiology and cardiology with the University of British Columbia. Dr Leipsic is also a Canada Research Chair in advanced cardiopulmonary imaging. He has over 480 peer-reviewed manuscripts in press or in print and over 300 scientific abstracts and is editor of two textbooks. He speaks internationally on a number of cardiopulmonary imaging topics with over 150 invited lectures in the last 4 years.

The highest activity a human being can attain is learning for understanding, because to understand is to be free.

– Baruch Spinoza

In the spring of 2019, two large randomized trials were published that showed transcatheter aortic valve replacement (TAVR) to yield similar-to-better outcomes in patients deemed low risk for surgery (1,2). These landmark studies have driven many in the field to suggest that all patients should be offered TAVR, given the equipoise or superiority in low-risk patients. This thinking ignores the role that CT angiography played in the screening of participants and the determination of appropriateness based on specific anatomic findings at CT, such as severe subannular calcification, adverse root features, low coronary height, and narrow (or shallow) Sinus of Valsalva width. What is more evident from these trials is that surgical risk is not particularly relevant to TAVR and we need to continue to learn about the drivers of risk in TAVR to better guide sizing and treatment decision making.

In this issue of Radiology: Cardiothoracic Imaging, Willemink et al (3) sought to retrospectively examine whether quantification of aortomitral continuity calcification (AMCC) as a whole—from aortic valve to mitral valve and including the aortic annulus, left ventricular outflow tract (LVOT), intervalvular fibrosa, and mitral annulus in between—is independently associated with all-cause 1-year mortality (primary outcome measure) or paravalvular leak or prolonged hospital stay (secondary outcome measures). In this study, a greater burden of AMCC was independently predictive of 1-year all-cause mortality irrespective of the Society of Thoracic Surgeons (STS) score. These findings suggest that severe AMCC should not be overlooked when evaluating patients for TAVR and may be considered in future risk stratification models as severe AMCC, as mitral annular calcification (MAC) is a marker of more advanced disease and atherosclerotic burden and contributes to increased medium-term mortality in this subset of patients after TAVR. Indeed, in a retrospective multivariable analysis, severe MAC was found to be strongly independently predictive of overall mortality following TAVR (hazard ratio [HR], 1.95; 95% confidence interval [CI]: 1.24, 3.07; P = .004), as well as of cardiovascular mortality (HR, 2.35; 95% CI: 1.19, 4.66; P = .01) and new permanent pacemaker implantation after TAVR (odds ratio, 2.83; 95% CI: 1.08, 7.47; P = .03) (4).

As we move forward with broader TAVR adoption, better understanding of not only procedural risk but also prognosis is paramount. The measure proposed in this analysis is attractive because it is simple to measure, making it highly reproducible, thus facilitating broad adoption. Prior to this, however, there are some outstanding questions that need to be answered. First, at present, the findings are merely hypothesis generating and would require validation in additional cohorts beyond this derivation population. Second, there were confounding variables, not adjusted for, that were significantly different between the groups, including pulmonary hypertension, coronary calcium, and corrected gas transfer. In addition, the nonrandomized and retrospective nature of the study augments the potential for uneven distribution of hidden confounders. Finally, whether this measure is either common or predictive in a lower risk population remains unclear and will require future investigation.

Even if these findings are confirmed, the field will be left with questions as to why AMCC is such a powerful prognosticator. Does calcium distribution really matter? We have seen prior publications that have noted incremental risk of rupture on the basis of subannular calcification and even variable risk based on the distribution (5,6). These prior analyses suggested that only aortic annular and high LVOT calcifications alter the characteristics of the landing zone (the valve cusps, aortic annulus, and LVOT—the area where a prosthetic valve sits when implanted) and are therefore likely to be the main drivers of procedural risk and complications, as opposed to calcification of the intervalvular fibrosa or MAC, which may just represent a marker of more advanced disease including mitral valve disease or atherosclerotic burden. Is this a measure that helps us better identify patients who may not experience improvement in mitral regurgitation? Clearly all of these potential mechanisms are possible and there are undoubtedly many more. This is, in many ways, the value of this assessment—to highlight the opportunities to extract more information from CT datasets that are acquired for the purpose of TAVR planning. While this analysis and the measurements that supported it were manually performed, the future undoubtedly lies in unstructured machine learning algorithms and texture analyses with radiomics which may allow for the identification and validation of other prognostically important anatomic variables.