Osama M Abdullah1, MarJanna Dahl2, Gavin Yeip1, Julia Cortino1, Arnold David Gomez3, Thomas Seidel1, Frank Sachse1, Kurt Albertine2, and Edward W Hsu1
1Bioengineering, University of Utah, Salt Lake ity, UT, United States, 2Pediatric Neonatology, University of Utah, Salt Lake ity, UT, United States, 3Electrical and Computer Engineering, Johns Hopkins, Baltimore, MD, United States
Synopsis
Diffusion tensor imaging (DTI) has emerged as the method of
choice for noninvasive quantifications of myocardial microstructure. However,
the origins and behaviors of DTI measurements as functions of myocardial
remodeling during development remain poorly understood. In this work,
conventional and bi-compartmental DTI and histological correlation were
performed on an animal model of myocardial development to investigate the
effects of tissue remodeling. Results indicate that tissue remodeling during
development manifests in progressively increased DTI transverse diffusivities, decreased
fractional anisotropy, and unchanged fiber orientation. The findings show that
DTI can be used to noninvasively characterize microstructural remodeling of the
myocardium during development.Introduction
Diffusion tensor imaging (DTI) (1) has emerged as the method of choice
for noninvasive quantifications of myocardial microstructure, which play an essential
role in the structural and functional properties of the heart. However, the origins and behaviors of DTI measurements
as functions of myocardial remodeling (e.g., myocytes growth) during development
and in disease remain poorly known or understood. In this work, conventional
and bi-compartmental DTI and direct histological correlation were performed on
an animal model of myocardial development (2) to investigate the effects of tissue
remodeling.
Methods
Hearts were obtained from 120 d and 130 d post-gestation,
full term (150 d) at birth, and term-born 5 month-old lambs (n=6-8 each),
arrested at cardiac diastole and fixed by perfusion using KCl and formalin as
described previously (3). DTI was conducted on a Bruker Biospec
7T scanner using multi-slice diffusion-weighted spin echo (SE) sequence (1500
ms TR, 30 ms TE, 64x64 matrix, 2.0 mm thickness, 4-7 cm
2 FOV depending
on heart size, 12 encoding directions with 5 and 15 ms pulse width and
separation, 1500 s/mm
2 b-value). Signal to noise ratio was above 80
for all B0 scans. Diffusion tensors were estimated as described previously (4), and the ranked diffusivities D
1,
D
2, D
3, their associated mean diffusivity (MD) and fractional
anisotropy (FA), and transmural span of the helix angle (HA) were averaged over
manually segmented left ventricular regions. To elucidate the source of the DTI
measurements, bi-compartmental DTI experiments (5), consisting of 48 gradients
directions and 12 b-values ranging from 100-5000 s/mm
2 were conducted
in 5 hearts randomly selected from each 130 d and 5 mo groups. The scalar diffusivities
for each compartment, MD and FA in addition to the angular difference (Δα ) between the principal eigenvectors (EV1)
of the fast and slow compartments were obtained. Moreover, hearts from the same
groups were cryo-sectioned, stained with wheat germ agglutinin (WGA)
and DAPI and were imaged using a Leica SP8 confocal microscope, to highlight the
extracellular space (ECS) and cell nuclei, respectively (3). Nucleus density (per mm
2),
myocyte size (length and width), and myocyte volume fraction (VF) were quantified
in regions from 3 different transmural biopsies per heart. Bonferroni-corrected
ANOVA and unpaired t-tests were performed for statistical
analysis of the conventional and bi-compartmental DTI results, respectively, with P < 0.05 considered statistically significant.
Results
Maturation was associated
with progressively decreased FA and increased MD, whereas HA appears
qualitatively similar among the groups (Figure 1). Changes in scalar DTI
parameters were linked to preferential increases of the radial diffusivities D
2
and D
3 (Table 1). The HA transmural span was not statistically
different among the groups. Representative FA maps of bi-compartmental DTI (Fig.
2) show that maturation leads to lower FA in both fast and slow compartments. Moreover, bi-compartment DTI parameters
summarized in Table 2 revealed that 5 mo group exhibited higher transverse diffusivities
accompanied by lower FA in both compartments, higher VF of the fast compartment,
and higher Δα between
compartments. Analyses of confocal images (Fig. 2) from 3 sections in each
heart revealed that hearts in the 5 mo group had larger myocytes (105±7 µm length and 18±2 µm width, vs. 46±5 µm
length and 6±1 µm width), lower nucleus density (1802±209 vs. 6122±294 per mm
2 ), and higher intracellular VF (78±3% vs. 67±5%).
Discussion and Conclusions
The
results indicate that tissue remodeling during development manifests in
progressively increased DTI transverse diffusivities, decreased FA, and
unchanged fiber orientation. Bi-compartmental DTI analysis reveals that the
changes are mediated by similar changes of diffusivities in the
compartments. The VF of the fast
diffusion compartment numerically matches the intracellular VF from confocal
microscopy, in both the magnitude and behavior. Because myocyte growth results in higher transverse
diffusivities but unchanged longitudinal diffusivities in both compartments suggests
that restriction due to physical compartmentalization may not be the
predominant mechanism underlying myocardial DTI measurements. Nevertheless, the findings show that
DTI is sensitive, and can potentially be used to noninvasively characterize
microstructural remodeling of the myocardium during development.
Furthermore,
the data suggests a role for DTI to characterize cardiac remodeling in developing
hearts affected by persistent pulmonary hypertension, which is a common
condition in preterm births (2).
Acknowledgements
This work was supported
by NIH R01 HL092055. References
1. Basser et al. Biophys
J. 1994;66(1):259–67.
2. Albertine et al. Am
J Respir Crit Care Med. 1999;159(3):945–58.
3. Seidel et al. LNCS.
2013;7945:300–7.
4. Abdullah et al. NMR
Biomed. 2014;27(11):1378–86.
5. Hsu et al. Magn
Reson Med. 2001;45(6):1039–45.