The big (and beautiful) picture
4D flow MRI made a splash in the imaging world more than a decade ago when Michael Markl, PhD, then a post doc in radiology at Stanford University in California, teamed up with engineers to demonstrate the feasibility of time-resolved, 3D phase-contrast MRI of the heart (J Magn Reson Imaging 2003;17:499-506). The technique—3D plus time—was considered an improvement over 2D imaging because it provided complete spatial and temporal coverage of an area of interest, capturing all of the regions, phases and directions of blood flow. The volumetric technique allowed physicians to both visualize and quantify flow.
The heart was a logical target, says Markl, who is now a professor of biomedical engineering and radiology at Northwestern University and director of cardiovascular MRI research operations at the Feinberg School of Medicine in Chicago. “The easiest application is to start with the vessels and the largest compartment we have in the body,” he says.
Within the cardiac realm, congenital heart disease also seemed like a good fit. Abnormalities, whether from defects at birth or consequent interventions or both, may force blood to flow in unexpected directions and patterns. Doppler echocardiography can assess flow but it provides only a 2D snapshot of one-dimensional velocities. Conventional 2D phase-contrast MRI can be used to quantify flow volumes, but it requires patients to hold their breath and, because the imaging plane must be predetermined, the data will include only abnormal flow within those parameters. 4D flow MRI, on the other hand, allows patients to breathe freely and blood flow can be studied retrospectively through any planes of interest.
“To capture what is going on in these [congenital heart] diseases, you really need to quantify flow in multiple locations and this technique lends itself to do that,” Markl says.
The results are visually captivating. Using 4D flow MRI, researchers have been able to compare hemodynamics in healthy and diseased patients and produce colorful images and striking videos. Cardiologists, already trained in reading MRI and 2D and 3D echo scans, intuitively understand the imagery, Markl says.
Michael D. Hope, MD, co-director of the Vascular Imaging Research Center at the University of California, San Francisco, calls the visuals “a beautiful thing … where you see complex blood flow unfold over time.” When he was a medical student at Stanford in the early 2000s, Hope joined Markl’s research team and is a co-author on a 4D flow cardiovascular MR consensus statement that outlined potential clinical applications for the technique (J Cardiovasc Magn Reson 2015;17:22). The challenge now is transitioning 4D flow from a validated and accurate research tool into clinical practice.
To succeed, proponents need to show its practical value above existing techniques, Hope says. “Part of the problem with advanced imaging is you can see things that are great, but if the information is redundant compared to the more standard [approaches], why would you bother with the effort of putting together all the complex images?”
4D flow has its drawbacks. A scan can take five to 15 minutes and, if the case is complex, it may be as long as 30 minutes, Markl says. That’s an improvement over its earlier days but still longer than the relatively short time required for a conventional MRI. 4D flow may be advantageous, though, Hope says, if multiple 2D scans are required because then the technologist must repeatedly readjust the scanner and reinstruct the patient about breath holds.
In the clinical setting, physicians expect results from scans to be available quickly, which presently is not the case with 4D flow, Markl says. “The data analysis is not terribly complicated; it is just not well integrated. The reason is that it has been developed at academic institutions and our focus is more on research questions. … We are not software developers who generate easy-to-use user interfaces that can be used clinically.”
Building a case for 4D flow
Congenital heart disease is the most common birth defect in the U.S., and the bicuspid aortic valve is considered the most common abnormality, with prevalence between 1 and 2 percent of the general population. “For guys like me,” says Patrick M. McCarthy, MD, chief of cardiac surgery and executive director at the Bluhm Cardiovascular Institute at Northwestern, “that is quite a bit of prevalence.” McCarthy also is a research collaborator with Markl.
Markl, Hope and McCarthy see a potential role for 4D flow as a tool for risk-stratifying and managing patients with bicuspid aortic valves for related complications, such as aortic dissection and aneurysm. Current recommendations for patients with bicuspid aortic valves and severe aortic enlargement recommend surgery for aortic dilation based on the diameter of the aortic root or ascending aorta (J Am Coll Cardiol 2016;67:724-31).
Proponents think 4D flow may offer a more nuanced assessment of risk. “Maybe there is a better way than to look at one number,” McCarthy says, for instance, by adding functional and physiological factors.
Bicuspid aortic valves are heterogeneous, and how blood flows, where and its velocity may vary with differing consequences. In a comparison using 4D flow MRI on healthy volunteers, patients with tricuspid aortic valves and patients with bicuspid aortic valves, Hope and colleagues showed that different leaflet fusions resulted in different flow jet directions (Radiology 2010:255:53-61). They observed abnormal helical blood flow patterns in patients with bicuspid aortic values, which were associated with eccentric flow jets in the proximal ascending aorta. They suggested that characterizing eccentric flow jets may help identify those at risk of aortic aneurysm.
Quantifying wall shear stress, or the frictional flow of blood on the aortic wall, using 4D flow MRI may also provide a prognostic marker for surgical intervention in patients with bicuspid aortic valves. In a study that compared hemodynamic forces exerted on the ascending aorta in patients with the most common bicuspid aortic valve morphology and three control groups (healthy young volunteers, age-appropriate participants with tricuspid valves and participants with tricuspid valves and aortic aneurysms), Markl and colleagues determined that wall shear stress was elevated at focal points in the ascending aorta in the bicuspid valve group compared with controls (Circ Cardiovasc Imaging 2012;5:457-66).
McCarthy, Markl and others have continued exploring wall shear stress with bicuspid aortic valve aortopathy, including how valve-related hemodynamics may influence the progression of disease. In a 4D flow study of the relationship between wall shear stress and remodeling in patients with bicuspid aortic valves, they found areas of increased wall shear stress had more markers of degradation based on tissue samples from patients who underwent ascending aortic surgery (J Am Coll Cardiol 2015;66:892-900). The results (see illustration above), if validated, may support the use of wall shear stress as a biomarker and help individualize patient care.
Going for the guidelines
McCarthy, Markl and the other members of their multidisciplinary research group now are focusing on making the cardiovascular community aware of 4D flow MRI and its potential, and building an evidence base for including 4D flow in clinical guidelines. Northwestern’s hospitals use 4D flow clinically and their expanding database includes a patient population that exceeds 500. They plan to conduct larger studies and follow-up analyses to assess 4D flow’s prognostic value.
“I want to make sure we get [patients] back for follow-up to see what is happening and to start looking to see if wall shear stress is correlating with changes in the aorta,” says McCarthy, who estimates that 4D flow will be clinically useful within the next two to four years.
He doesn’t anticipate reimbursement to be a barrier, since funders already pay for MRI for congenital heart disease. The technique requires additional software and training for the technologist but no changes to scanners, Markl says. Industry has come onboard, with a few vendors now offering commercial software.
Industry’s participation will help further standardize analytical tools and facilitate uptake, says Markl, who started his career as a medical physicist. “The one thing that is absolutely critical for these new techniques is that it is not just radiology,” he adds. “There needs to be an integrated effort by physics, engineering, computer science, cardiology, radiology and industry to get this off the ground. It is always a process that is slower than you want it to be, but we have made great progress and we are working with all sides to make that happen.”