How Well Does CT-derived Fractional Flow Reserve Predict Outcome?

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

Timothy A. Fairbairn, MBChB, FRCP, PhD, is a consultant cardiologist subspecializing in cardiovascular imaging at Liverpool Heart and Chest Hospital. He practices in advanced echocardiography, cardiac MRI, and cardiac CT. His clinical and research interests include coronary artery disease, aortic valve disease, and inherited cardiac conditions. He has been an advocate for the introduction of cardiac CT angiography as the first-line diagnostic test for stable chest pain in the United Kingdom and is a member of the NICE Adoption and Impact Programme Reference Panel for new technologies.

Russell Bull, BSc, MBBChir, MRCP, is a consultant cardiac radiologist at Royal Bournemouth Hospital. He specializes in cardiac CT and cardiac MRI.

Introduction

The assessment and management of symptomatic patients suspected of having coronary artery disease (CAD) remains a challenge. A debate rages on among cardiologists as to whether it is the anatomic degree of stenosis caused by coronary atheroma or the presence of myocardial ischemia (physiology) that matters most. An accurate diagnosis of angina and appropriate medical treatment would appear to be the most important issues to establish, but the choice of who and when to refer to the catheterization laboratory remains the most contentious and discussed management decision. Most modern-day stable chest pain studies have shown that the risk of an adverse cardiovascular event occurring, such as myocardial infarction or cardiac death, is actually relatively low at less than 1% annual event rate (1). This is below the threshold that most guidelines use as a cutoff to recommend invasive coronary angiography (ICA) and consideration of revascularization. Thus, it is reasonable to argue that a diagnosis and symptom-driven management is all that is required and that the debate is a false one, reserved for those patients with failed medical treatment. Yet, cardiologists are filled with the understandable fear of missing the patient who may benefit from testing and revascularization on the basis of prognosis. If testing is required, then the optimal test should be one that helps this decision process and guides patient management as to who will benefit the most from ICA and revascularization based on prognostic grounds.

Cardiac CT angiography is becoming the preferred first-line test for the assessment of suspected CAD. Its strength lies in the high sensitivity for the detection of CAD and excellent long-term prognostic data in the absence of CAD. Cardiac CT angiography has also been shown to have the ability to determine prognosis based on anatomy alone related to the number of coronary arteries diseased and whether the CAD is nonobstructive (<50% degree stenosis) or obstructive (≥50% degree stenosis) (2). However, cardiac CT angiography is limited by its inability to demonstrate ischemia, as the presence of ischemia has been shown to provide incremental prognostic information and importantly helps guide revascularization with improved outcomes compared with anatomic assessment alone (3).

The arrival of fractional flow reserve derived from coronary CT (FFR_CT) as a test that can assess anatomic disease severity, as well as ischemia noninvasively, has the potential to change clinical practice and end the debate. FFR_CT uses the thin-section data of the acquired cardiac CT angiography, advanced computational flow dynamics, Newtonian physics, and deep learning to produce a three-dimensional model of the coronary arteries with a per-vessel and per-stenosis map of myocardial ischemia. Several studies have shown that FFR_CT has a good correlation to invasive measures of ischemia (invasive FFR) (4), outperforms some conventional methods of ischemia testing (5), and changes patient management in two out of three cases compared with cardiac CT angiography alone (6). These advantages over cardiac CT angiography alone would appear to translate into improved resource management, as inferred by higher rates of revascularization at the catheterization laboratory. Whether these increased rates of revascularization translate into better patient outcomes has remained less well established.

In this issue of Radiology: Cardiothoracic Imaging, McNabney et al (7) add to our understanding of the ability of FFR_CT to predict clinical outcomes. While limited by its retrospective, single-center nature, the reporting of 200 sequential FFR_CT cases' midterm (median, 477 days) outcomes is important when the long-term evidence base remains small in the early phase of this new technology.

The likelihood of a positive FFR_CT, as expected increased incrementally with higher degree stenosis as shown by the Coronary Artery Disease Reporting and Data System (CAD-RADS) (CAD-RADS 2, 18.8%; CAD-RADS 3, 30.6%; and CAD-RADS 4, 53.5%). A more interesting observation was the low incidence (9.5%) of anatomically insignificant stenoses causing a “gray zone” FFR_CT result when using a stenosis-specific rather than distal vessel interpretation. It has been a matter of concern that diffuse disease was resulting in these gray-zone results, leading to false-positive findings with reduced diagnostic accuracy of the test. This article would suggest that essential to the test accuracy is the method of reading the result.

How an FFR_CT result is interpreted and used is key to understanding the test itself. The authors choose not to just report stenosis physiologic significance as positive (FFR_CT positive) or negative (FFR_CT negative) but also subclassify according to anatomic significance (significant stenosis positive ≥ 50% or significant stenosis negative < 50%). This allows us the opportunity to observe patterns of behavior in a single center using FFR_CT. What is most revealing is that while the frequency of ICA and revascularization increased with an FFR_CT positive result (63.5% and 41%, respectively), an FFR_CT negative result still led to ICA in 30% and revascularization in 15% of individuals, all of whom had an anatomically obstructive stenosis. While there is no mention of the degree of medical treatment in these patients, these findings are consistent with other studies (6) and suggest that while an FFR_CT negative result portends a good outlook, it appears difficult for the interventional cardiologist to ignore the anatomy.

The primary importance for any new test is safety. McNabney et al build on the safety profile set out in ADVANCE (Assessing Diagnostic Value of Noninvasive FFR_CT in Coronary Care) (6) and from the Aarhus group (8) that a negative FFR_CT (>0.80) is a good short-to-midterm prognostic marker with no events occurring in this group. However, as the rates of adverse events in contemporary stable CAD cohorts are low, this may be falsely reassuring. In addition, no events occurred in patients with nonobstructive (<50%) stenosis, therefore what does FFR_CT further add to our patient management? This is particularly relevant when you take into account that ADVANCE just failed (P = .06) to show a difference in events between the FFR_CT positive and FFR_CT negative cohorts at 1 year (9).

Perhaps more important than a negative result is the ability of a positive one to predict adverse outcomes. McNabney et al use the vessel-oriented clinical event (VOCE) of late revascularization (>90 days), myocardial infarction, or cardiac death. These are well-established end points used in clinical studies. No deaths occurred and only four myocardial infarctions, as late revascularization was the main VOCE observed (n = 22, 85%). FFR_CT positive and stenosis severity greater than 70% (but not 50%) were independent predictors of VOCE on multivariable analysis. While six (23%) late revascularizations occurred in FFR_CT negative patients, all of whom had an anatomic obstructive stenosis (≥50%), there was no independent invasive assessment of ischemia to adjudicate. These patients thus had anatomic-driven management, which runs contrary to the guidelines. Importantly, two patients who underwent bypass surgery had a left main stem (LMS) stenosis and diffuse three-vessel CAD with distal vessel tip positivity but no stenosis-specific positivity. This brings us back to how the result is interpreted, as stenosis-specific values likely increase test specificity but lower the sensitivity. This form of reporting should thus be used cautiously in instances of known high-anatomic risk (LMS or three-vessel disease).

The granularity of this data should be congratulated as it allows us to develop a greater understanding of the impact of anatomy and physiology on prognosis. A difficulty persists in determining whether FFR_CT can be used to predict better outcomes with revascularization when other studies have failed to show this superiority over medical therapy alone. If we have learned anything it is that both anatomy and physiology matter, as the individuals at greatest risk will be those who have high-risk anatomic and physiologic results. It is likely that FFR_CT “physiology” does not simply trump CT angiography “anatomy” but the incremental information afforded by both tests can be used to provide better prognostic data, improve management decisions, and potentially find the holy grail of improved clinical outcomes. It is in this realm that CT angiography and FFR_CT have the potential to answer the cardiologist's dilemma.