Abnormal 3-D MRI Flow Patterns in Adolescents Patients With Bicuspid Aortic Valve

Introduction-Background The reported incidence of bicuspid aortic valve (BAV) based largely
on autopsy studies varies between 0.4% to 2.25% of the population. At least a third of
patients with a bicuspid aortic valve will eventually develop complications which include
aortic stenosis, aortic regurgitation, infective endocarditis, aortic dilatation, aortic
aneurysm formation, or aortic dissection. Given that in combination all other forms of
congenital heart disease are thought to be present in 0.8% of live births, Ward has
suggested that bicuspid aortic valves are likely to result in more morbidity and possibly
mortality than the effects of all other congenital heart defects combined. Since many
patients with bicuspid aortic valves do not develop significant problems until well into
adulthood, clinicians have not focused much attention on the "normally" functioning bicuspid
aortic valve. However, at a mean age of 17.8 years 52% of males with normally functioning
aortic valves already have aortic dilatation, implying that associated aortic stenosis
and/or regurgitation is not an obligate precursor.

Previous studies have demonstrated an association between bicuspid aortic valves and
dilatation of the aorta. Gurvitz et al. performed echocardiographic measurements of the
aortas of children and found that those with bicuspid aortic valves had significantly larger
aortas compared with controls, regardless of the presence of aortic stenosis or
regurgitation. Nevertheless, the definition of clinical predictors of the risk for the
development of aortic dilatation/dissection and an understanding of the mechanisms leading
to aortic dilatation are lacking. Specifically, it remains unclear as yet whether such
dilatation is secondary to abnormal flow patterns and shear stresses resulting from the
bicuspid valve morphology or whether it is a manifestation of a distinct underlying
structural problem with not only the aortic valve but also the aortic root including the
ascending aorta.

Fernandes et al. recently reviewed the echocardiograms of 1,135 children with bicuspid
aortic valves and found that there were differences among the patients related to aortic
valve morphology. For instance, associated moderate or greater aortic stenosis was present
in 9.7% of patients with fusion of the intercoronary commissure of the aortic valve, vs
25.9% of patients with fusion of the right-coronary and non-coronary commissure, and in none
of the patients with fusion of the left coronary and non-coronary aortic valve leaflets.
Moreover, fusion of the right coronary and non-coronary cusps resulted in a two-fold higher
risk of at least moderate aortic regurgitation compared with the other types of bicuspid
aortic valves. However, this generalized division of bicuspid aortic valves into three
types based on which commissure is fused is likely an oversimplification, and Fernandes did
not report on the relationship between valve morphology and aortic root dilatation.
However, Novaro et al. found that adult patients (mean age of 54 year) with fusion of the
right/non-coronary commissure tended to have larger mid-ascending aortas compared with
patients with fusion of the intercoronary commissure, but the difference did not reach
statistical significance. Nevertheless, both personal clinical observations and recent
publications support the assertion that individual bicuspid aortic valves may function quite
differently from one another. For instance, there are bicuspid aortic valves in which there
are two nearly symmetric leaflets whereas others may have a dominant cusp to varying
degrees, resulting in a markedly eccentric orifice when the valve opens. In an elegant
experimental model using excised bicuspid valves analyzed with intravascular ultrasound and
high-resolution videography, Robicsek et al demonstrated that asymmetric bicuspid valves
"induced extensive recirculation vortices in the ascending aorta." They found that the
vortex was not "trapped" in the sinuses of Valsalva as it is in a normal tricuspid aortic
valve, but instead is extended into the ascending aorta and aimed toward the right
anterolateral aspect of the aorta (the convexity of the aorta). This correlates with
published reports and our own personal experience in performing clinical cardiac magnetic
resonance imaging (CMR) studies on patients with bicuspid aortic valves and aortic root
dilatation in whom there is oft-noted asymmetry in the pattern of dilatation of the
ascending aorta and resulting in an oval-shaped rather than a circular aortic root and
ascending aorta.

Extrapolating from the studies of Robicsek, variations present in the geometry of bicuspid
valves are likely to result in varying degrees of turbulence even in the absence aortic
valve stenosis. In theory, the turbulent blood flow directed at a particular segment of the
aortic wall may result in local changes to the aortic wall leading to asymmetric aortic
dilatation. These changes are postulated to result from receptors present on both
endothelial and smooth muscle cells that have the ability to adapt to sheer stress by
altering local gene expression which in turn modulates the tension between focal adhesion
sites, integrins, and the extracellular matrix. Moreover, fibrillin-1, which is deficient
in patients with dilated aortic roots and Marfan syndrome, has also been found to be
deficient in the aortas of patients with bicuspid aortic valves. Similarly, Cotrufo et al.
found asymmetric patterns of matrix protein expression and content and asymmetric patterns
of elastic medial wall degeneration in the aortic walls of patients with bicuspid aortic
valves. In addition, these investigators found differences between patients with bicuspid
aortic valves and associated stenosis vs regurgitation, suggesting that focally-directed
aortic turbulence may be influencing local gene expression and potentially explaining the
asymmetric pattern of aortic dilatation that is often observed. Lastly, several recent
studies have highlighted the relationship between circulating plasma MMP-2 and MMP-9 levels
either in aortic dilatation in adult patients with thoracic or abdominal aortic aneurysms or
in adults with aortic dilatation secondary to systemic hypertension. In addition, histologic
examination of aortic aneurysm tissue in adult patients with bicuspid aortic valves
demonstrated increased MMP-2 expression vs. aortic aneurysms in adults with trileaflet
aortic valves.

Recently investigators have used cardiac MRI (CMR) to study flow patterns in the aorta. CMR
is a noninvasive technique that permits 3-dimensional anatomic characterization as well as
assessment of aortic flow. Markl et al. have used time-resolved 3-dimensional phase-contrast
magnetic resonance imaging (3D-PCMRI) techniques to characterize aortic flow in both normal
adults and in adults after aortic root replacement. These authors established that
3-dimensional magnetic resonance velocity mapping was a useful technique to visualize and
qualitatively assess the complex aortic flow patterns in both normal controls and in
patients with aortic pathology. Similarly, Kvitting et al., who studied both adult
volunteers and 2 patients with Marfan syndrome following aortic valve sparing operations,
found that the alterations to the normal aortic sinus architecture following surgery
resulted in a loss of the normal vortical flow patterns seen in the aortic sinuses.

Specific Aims

1. Develop clinically feasible MRI imaging strategies to perform a complete evaluation of
structure and functional dynamics of the heart, including the aortic valve, aortic
root, and ascending aorta, as well as full volumetric 3-D ciné flow velocity encoding
of the entire aortic root region and proximal ascending aorta.

2. Define qualitative and quantitative metrics to characterize different types of bicuspid
aortic valve morphology and flow patterns. These metrics will be based on structural
information and the blood flow patterns in the aortic root, observed in MRI studies of
patients with diagnosed bicuspid aortic valves, as compared to observation in control
subjects with known normal aortic valves and anatomy of ascending aorta.

3. Correlate different bicuspid aortic valve morphologies and observed characteristic flow
patterns with the risk of development of aortic dilatation at various levels of the
aorta, such as the aortic root, proximal ascending aorta, and distal ascending aorta.

4. Quantify plasma concentrations of matrix metalloproteinase-2 (MMP-2) and matrix
metalloproteinase-9 (MMP-9) levels as well as the levels of their specific tissue
inhibitors, respectively tissue inhibitor of metalloproteinase-2 (TIMP-2) and tissue
inhibitor of metalloproteinase-1 (TIMP-1) in controls and bicuspid aortic valve
patients.

5. Correlate biochemical markers of extracellular matrix degradation (both MMP and TIMP)
with the presence or absence of aortic dilatation in both controls and in patients with
bicuspid aortic valves.

Study Design We will prospectively perform cardiac magnetic resonance studies on 30
patients, ages 10-18 years with known bicuspid aortic valves (BAV), and on 15 normal control
subjects. The BAV patient population will consist of 15 patients with aortic root
dilatation for body surface area as determined by published pediatric echocardiographic
normal values and 15 without aortic dilatation. Echocardiographic normals indexed for body
surface area will be used, as there are no published normal CMR values for aortic dimensions
in pediatric patients using cine MR techniques; however aortic measurements by
echocardiography and CMR in a pediatric patient population have been shown to reveal no
significant differences except for the descending aorta. Since either significant aortic
stenosis or regurgitation has been associated with aortic dilatation, patients will be
excluded if they have more than mild stenosis or regurgitation.

As in previous published reports, mild aortic stenosis will be classified as a mean gradient
across the aortic valve <17 mmHg on routine clinical echocardiography. Mild aortic
regurgitation will be defined on routine clinical echocardiography as resulting in a left
ventricular end-diastolic dimension <2 standard deviations below the mean for body surface
area and the absence of pan-diastolic retrograde flow on pulsed Doppler evaluation of the
abdominal descending aorta. Studies will be performed without sedation, but with
breath-holding instructions as appropriate.

Controls will consist of 15 patients with normal aortic valves and aortas who are undergoing
clinically indicated CMR studies for reasons such as chest pain or assessment for
cardiomyopathy and who are found to have structurally/functionally normal hearts.

MRI Methods Cardiac magnetic resonance (CMR) studies will be performed in a GE 1.5 T MR
system with Release 12M4 cardiovascular software. MR imaging will focus on of
visualization in high definition of the anatomy of the aortic valve and aortic root
utilizing 2-dimensional and 3-dimensional FIESTA cine MR sequences. Flow data will be
acquired using 2-dimensional, or, whenever allowed within imaging time constraints, fully
3-dimensional phase velocity mapping sequences with full vector-encoding of velocities at
200 cm/s maximum velocity encoding VENC, adjusted up if velocity aliasing is observed. For
2-D flow acquisitions in planes across the main flow direction the in-plane directional
components may be acquired at lower velocity encoding to achieve higher accuracy. To
resolve blood flow and cardiovascular anatomy as a function of the cardiac phase (cine
imaging), image acquisition will be synchronized with the vector cardiogram to reconstruct
approximately 16-20 cine images representing the different phases of the cardiac cycle.
Initial data acquisition for structural and velocity-encoded imaging will be performed with
standard pulse sequence software as supplied by General Electric with the imaging device.
These scans will initially be acquired in free-breathing mode. A series of 5-10 studies on
normal volunteers will be conducted to optimize the protocol within those constraints, and
develop guidelines for an prescribed add-on imaging protocol, ideally limited to not more
than 20 minutes of scan time, for acquisition of the aortic scans. Priority solutions for
respiratory motion compensation during the acquisition depending on the demonstrated
reliability of the available options (free-breathing, breath-holding, respiratory k-space
ordering, navigator- or belt & bellows-monitored respiratory gating) in these volunteer test
studies, and based on the capability of each study subject to accommodate these options.
The scans defined in this protocol will be added initially to the clinical protocol of all
subjects in the study. In parallel with the protocol development and implementation by
standards pulse sequence methods, we will pursue a collaboration effort with the MRI imaging
resource center at Stanford University. The Stanford group has developed a highly optimized
image acquisition method which allows fully vector-encoded 3-D respiration-ordered
acquisition of an arbitrarily positioned rectangular 3-D imaging volume. We anticipate that
this acquisition software, which will require running specialized pulse sequence software
and off-line image reconstruction on one of our computer workstations, will improve the
coverage and overall resolution of the flow-encoded data significantly. We have the
required expertise and research software tools available from General Electric, but
currently only enabled for research on the WCI scanner at Emory. For this project, but
possibly concurrently with transition of software development efforts for other research
projects (Noquist project) we will request authorization from GE to run the required
research patches on the Egleston 1.5T GE instrument.

Plasma MMP and TIMP quantification methods Blood samples will be taken from peripheral veins
by nursing staff.MMP and TIMP levels will then be measured using commercially available
sandwich enzyme-linked immunosorbent assay (ELISA) kits which have previously been validated
for use in human tissue homogenates.Serum MMP levels will be determined using MMP-2 and
MMP-9 monoclonal antibodies.Levels of TIMP-1 and -2 will be determined using monoclonal
antibodies.

Image Processing Methods Visualization and quantitative analysis of velocity mapping data
will be performed, initially exclusively, utilizing in-house HeartViz software developed by
Marijn Brummer, Ph.D.The current version of the software allows fully interactive 3-D
dynamic integrated display of structural information through texture-mapped slice display
and time-resolved volume rendering, in conjunction with dynamic flow vector field
visualization from multiple flow acquisitions in the same view. This software has been
developed by the investigators during the past ten years, and has been proposed and used for
evaluation of a variety of congenital and acquired cardiac diseases, including aortic
coarctations, VSD, aortic dissections, and peri-atrial tumors.