Heart and Metabolism

Identification of patients at risk for left ventricular remodeling: role of magnetic resonance tissue tagging and strain analysis

Marco J.W. Götte¹, Joost P.A. Kuijer², J. Tim Marcus², Albert C. van Rossum¹
¹Department of Cardiology, ²Department of Clinical Physics and Informatics, VU University Medical Center and Institute for Cardiovascular Research ICaR-VU, Amsterdam, The Netherlands
Correspondence: Dr M.J.W. Götte, Department of Cardiology, VU University Medical Center, P.O. Box 7057, 1007 MB Amsterdam, The Netherlands.
Tel: +31 20 444 2244, fax +31 20 444 2446, e-mail: mjw.gotte@vumc.nl

Introduction
Early identification of patients at risk for left ventricular remodeling after myocardial infarction will enable therapy to be adjusted before significant remodeling occurs. This may contribute to the prevention of heart failure. Metabolic imaging techniques using SPECT or PET are widely used for assessing myocardial ischemia and viability,[1] and for identifying patients who may potentially benefit from revascularization.[2] Most studies using these techniques, however, are carried out in patients with already severely depressed left ventricular function, and although they provide important data with respect to prognosis and patient management, they do not address the identification of patients prone to remodeling.
The remodeling process, characterized by progressive dilatation and impaired global function, is attended by regional differences in myocardial function and mechanical loading, both important stimuli that contribute to the progression of the postinfarct ventricle toward endstage heart failure.[3–7]
Magnetic resonance (MR) imaging has emerged as an important imaging modality for accurate assessment of global ventricular function.[8] MR tissue tagging offers the unique feature of noninvasively quantifying regional, intramural myocardial function in vivo.[9–15] This imaging modality thus enables study of the relationship between both global and regional function after infarction.
This article will discuss the potential value of MR tissue tagging and strain analysis for the detection of patients at risk for ventricular remodeling.

Mechanical implications of ventricular remodeling
Left ventricular remodeling has been suggested to play an important role in the progression of heart failure in patients with ventricular dysfunction,[16] and therefore contributes to the increased morbidity and mortality after myocardial infarction.[17] Serial follow-up studies of left ventricular dimensions and regional myocardial function after infarction have demonstrated that ventricular remodeling is attended by an early[18–20] and persistent[4] difference in segmental performance of the myocardium. In the early phase after infarction, adaptive responses are initiated in the noninfarcted remote myocardium to preserve stroke volume. Infarct expansion causes deformation of the infarct border zone and remote myocardium. This change in geometry alters the Frank-Starling relationship and augments shortening.[21] As a result, the remote myocardium becomes hyperkinetic and a — temporary — circulatory compensation is achieved.
Infarct expansion in the acute phase after myocardial infarction results in wall thinning and ventricular dilatation and causes elevation of wall stress, a major determinant of ventricular performance and an important stimulus for hypertrophy.
Late after infarction, alterations in myocardial wall architecture and ventricular geometry develop to distribute the increased wall stress more evenly. The rearrangement of bundles (‘slippage’) of myocytes[22] and myocyte hypertrophy[23] in response to the elevated wall stress cause a reduction in function of the remote myocardium and progressive impairment of global ventricular performance.
These findings suggest that besides humoral factors,[18,24,25] functional and mechanical factors play an important role in the process of remodeling,[3,5–7] and support the concept that remodeling is a regional process, induced by the local myocardial damage due to infarction, which involves the entire ventricle as more and more contractile units become exposed to the damaging effects of mechanical overload.
To improve understanding of the interaction between regional and global function in relationship to postinfarction remodeling, accurate quantification of regional myocardial function in different parts of the ventricular wall and changes in ventricular geometry are required.

MR tissue tagging
Until a decade ago, because of the need for identifiable landmarks or material points within the ventricular wall which can be followed during contraction, quantification of myocardial deformation was restricted to animal experiments and patients undergoing coronary angiography or cardiac surgery.
In 1988, Zerhouni et al[9] introduced MR tissue tagging, a method for noninvasive selective labeling of myocardial tissue. This was followed by the introduction of grid tagging, which provides a large number of traceable tags within the myocardium.[11,12]
The basic idea of grid tagging is to alter locally the magnetization properties of the tissue prior to image acquisition.

Figure 1. MR cine images (top row) and MR tissue tagging images (bottom row) at end-diastole (left) and end-systole (right). The tag line distance is 7 mm. The cine images demonstrate wall thickening during contraction, but intramural deformation is not visualized. In the tagged images, the tags are a temporary property of the myocardium and therefore the tag lines move along with the tissue. The deformed tag lines visualize directly the underlying motion of the myocardium.

The modulation of the magnetization appears as dark (‘tag’) lines on the images (Figure 1, bottom row). Because the magnetization is a property of the tissue, the tag lines move along with the tissue in which they are created (Figure 1, top right). The tag intersection points of the grid serve as intramural landmarks. The deformed tagging pattern reflects the underlying motion of the heart wall. By tracking the motion of the tag lines throughout the cardiac cycle, the intramural myocardial deformation can be quantified.[10]

Myocardial strain
The tag intersection points are used to define triangular elements of the myocardium across the heart wall (Figure 2). Homogeneous strain analysis is used to compute the deformation of each triangle.[12,13] The strain components — radial stretch (er) and circumferential shortening (ec) — are computed with respect to the radial and circumferential directions, respectively (Figure 2). Positive radial strains (er) represent the local contribution to wall thickening. Negative values for er imply local wall thinning. Negative circumferential strains (ec) quantify local circumferential shortening or myocardial contraction. Positive circumferential strains represent circumferential lengthening, associated with local dilatation.

Figure 2. On the tagged images, the white diamonds indicate the intersection points of the tagging grid. By using groups of three intersection points, multiple triangular elements of myocardium can be created. The variables r and c both represent a line segment in the undeformed state, in radial and circumferential direction, respectively. The variables r’ and c’ represent the same line segment in the deformed state.

Global ventricular function after infarction
Ventricular mass and volume can be measured accurately using cine MR imaging.[8] In the semiacute phase after infarction, the ratio of the enddiastolic volume and muscle mass reflecting the overall ventricular wall tension (Figure 3)[26] is equal in patients who demonstrate ventricular remodeling at follow-up and in patients without remodeling (0.99 ± 22 mL/g vs. 1.03 ± 0.13 mL/g, P = ns, respectively).

Figure 3. Ratio of end-diastolic volume to left ventricular mass (EDV/LVM), a reflection of global wall stress, at two different time points after infarction in patients with remodeled and nonremodeled ventricles.

These values are comparable to the normal value (0.9 mL/g) reported in the literature.[27] However, at 3 months’ follow-up, a significant increase in global wall stress (P < 0.05) was observed, mainly due to progressive ventricular dilatation (enddiastolic volume index from 84 ± 23 mL/m2 to 108 ± 26 mL/m2, P < 0.01), in patients with a poor ejection fraction (<40%). The global wall stress remained unchanged in patients with an ejection fraction >40%. These observations emphasize that wall stress is an important determinant of global ventricular function.

Regional myocardial function after infarction
At follow-up after infarction, changes in myocardial function are observed not only in the infarcted area but also in the noninfarcted adjacent and remote myocardium. In the semiacute phase after infarction, function in the infarcted area is severely depressed compared with that in the adjacent and remote regions (P < 0.001). The radial stretch (Figure 4A) in the infarcted area continues to fall during follow-up in patients with ventricles subject to remodeling. In addition, the radial stretch in the remote myocardium falls from 24 ± 7% to 16 ± 5% (P < 0.005).

Figure 4. Percent radial stretch (A) and circumferential shortening (B) in the infarct, adjacent and remote regions at 1 week and 3 months after myocardial infarction (MI), in patients with a remodeled ventricle at follow-up.

The same observations were made for circumferential shortening (Figure 4B). Shortening was severely impaired in the infarcted area (P < 0.001) compared with the adjacent and remote regions. During follow-up, no improvement in shortening was observed in the infarcted area, while a significant reduction in circumferential shortening in the remote region occurred (from –15 ± 5 to –12 ± 4%, P < 0.01).
Of both strain parameters, circumferential shortening in the semiacute phase after infarction was demonstrated to have the best predictive value for the development of ventricular remodeling. An increased value for circumferential shortening (less negative) in the remote area implicating reduced systolic shortening is associated with an increase in volumes and a decrease in ejection fraction during follow-up (regression coefficient obtained from generalized estimating equations: 2.525, P < 0.01).

Conclusion
Ventricular remodeling is associated with a poor long-term prognosis and can be considered a primary target for treatment. To assess the impact of ventricular shape on regional function and reverse, accurate noninvasive measurements of ventricular geometry and function are required.
MR imaging provides the unique feature of noninvasively quantifying regional, intramural myocardial function in vivo and offers even greater accuracy for assessment of global ventricular function. These features enable study of the relationship between both global and regional function, may provide new insights into the mechanical aspects of remodeling, and lead to reductions in sample size requirements for interventional studies.
Two-dimensional myocardial strain quantified in the semiacute phase after first infarction provides noninvasive data which are predictive of ventricular remodeling. Impaired shortening is associated with an increase in ventricular volumes at follow-up.

REFERENCES

1: Lancet 1998 Mar 14;351(9105):815-9