Background

Cardiovascular magnetic resonance (CMR) plays an increasingly important role in the diagnosis and assessment of coronary artery disease (CAD). It has evolved into a comprehensive clinical tool with the unique capability of assessing myocardial function; viability and perfusion in a single examination [1].

Wall motion analysis with CMR has a pivotal role in clinical practice. It is considered the gold standard for visualizing left ventricular (LV) endocardial wall motion at rest; as well as during low and high dose dobutamine stress to assess myocardial hibernation and ischemia. At the present time; image analysis is most commonly performed qualitatively. However diagnostic accuracy of qualitative assessment has been shown to be considerably operator dependant [2].

Deformation assessment of tagged lines within the myocardium may overcome these limitations however requires acquisition of additional tagging sequences and post processing [3]. Recently CMR myocardial feature tracking (FT); a technique analogous to echocardiographic speckle tracking; has been introduced [4]. CMR-FT allows tracking of tissue voxel motion of cine-CMR images with a potential to assess longitudinal; circumferential and radial myocardial strain as well as velocity; displacement and torsion independent of additional sequences. A good agreement of CMR-FT versus myocardial tagging with harmonic phase imaging (HARP) as a reference standard has been demonstrated [5]. However it is unclear; whether this approach would allow the response to dobutamine stress to be quantified [6]. The aim of the current study was to determine the ability of CMR-FT for quantitative wall motion assessment at rest and during intermediate dose dobutamine stress in healthy volunteers.

Methods

Ten healthy volunteers underwent CMR on a 1.5 Tesla scanner (Intera R 12.6.1.3; Philips Medical Systems; Best; The Netherlands). The study protocol was approved by the Institutional Review Board at the University of Nebraska Medical Center. All participants gave written informed consent.

Cardiovascular magnetic resonance

All CMR measurements were performed in the supine position using a 5-channel cardiac surface coil. LV dimensions and function were assessed with an ECG-gated steady state free-precession cine sequence during brief periods of breath-holding in the following planes: ventricular 2-chamber; 4-chamber; and 12 to 14 equidistant short-axis planes (slice thickness 6-8 mm; gap 0-2 mm) completely covering both ventricles. The field of view was 360 × 480 mm and matrix size 196 × 172. Dobutamine stress imaging was performed as previously described [7]. Repeat short-axis stacks were acquired with 10 and 20 μg · kg-1· min-1 of dobutamine; respectively.

Ventricular volumes and function

End-diastolic (EDV) and end-systolic volumes (ESV); stroke volume (SV); and ejection fraction (EF) were measured as previously described using commercially available software packages (View Forum; Philips) [8]. Ventricular volumes were adjusted to body surface area. All parameters were analysed at rest; 10 and 20 μg · kg-1· min-1 of dobutamine stress.

Feature tracking

CMR-FT analysis of strain was performed using a dedicated software prototype (Diogenes MRI; Tomtec; Germany). The 4-chamber view was used to calculate right ventricular (RV) and LV longitudinal strain and LV radial strain (EllRV and EllLV and ErrLAX) at rest. LV short axis circumferential (EccSAX) and radial strains (ErrSAX) were derived from a mid-ventricular short-axis view containing both papillary muscles. The RV upper septal insertion point of the LV was manually detected to allow accurate segmentation according to a recognized standard model [9]. Endocardial contours were manually drawn in all analyzed slices by one skilled observer (AS; 7 years of experience). EccSAX and ErrSAX were analysed at rest; 10 and 20 μg · kg-1· min-1 of dobutamine stress. A second observer (SK; 4 years of experience) re-analysed the images to assess inter-observer variability. The mid-ventricular short axis images analyzed by the second observer were at exactly the same slice position as for the first observer. The first observer repeated the measurements after a period of 4 weeks to assess intra-observer variability. Figure 1 shows a representative example of the tracking of LV and RV in the respective views.

Figure 1
figure 1

Tracking in Short-Axis and Long-Axis Orientation. The figure shows a representative example of the tracking in Short-Axis and Long-Axis Orientation of the left ventricle (LV) and the right ventricle (RV).

Full size image

Comparison with natural radial strain

Natural radial strain values were obtained as an external reference standard and compared to the respective CMR-FT ErrSAX values [10]. In brief end-diastolic and end-systolic wall-thickness (EDWT and ESWT) were quantified in identical segments as analysed for ErrSAX using commercially available software (Philips View Forum; The Netherlands) [11]. Natural radial strain values were calculated according to the following equation: log e (ESWT/EDWT) as previously validated [10]. 95% confidence intervalls of the difference and p-values were calculated to compare the 2 techniques [11].

Statistics

We have applied a paired t-test followed by a Bonferroni-Holm correction as a multiple test procedure to compare measurements at rest and with dobutamine stress after proving a normal distribution of the sample. Intra- and inter-observer variability analysis were performed using the method proposed by Bland and Altman [12]. A p-value of <0.05 was considered statistically significant. All data analysis was performed with PASW statistics for Mac 18.0.0 (SPSS Inc.; Chicago; Illinois; USA).

Results

The image quality was sufficient to perform strain analysis in all segments for all subjects. Gender was equally distributed and LV and RV volumes were within normal limits [13]. Participant demographics are shown in table 1. There were no side effects to dobutamine exposure. There was significant (p < 0.05) increase of heart rate; mean blood pressure and cardiac output between rest and both levels of dobutamine as well as between 10 and 20 μg · kg-1· min-1 of dobutamine

Table 1 Subject Characteristics
Full size table

Strain parameters at rest

Results at rest are displayed in table 1.

Dobutamine stress cardiovascular magnetic resonance

Changes in EccSAX and ErrSAX were significant between rest and both levels of dobutamine as well as between 10 and 20 μg · kg-1· min-1 of dobutamine (table 2; table 3; figure 2). In parallel with these changes LV-EF increased significantly with 10 and 20 μg · kg-1· min-1 of dobutamine (table 2; figure 2).

Table 2 The hemodynamic response and the response in strain parameters to 10 and 20 μg · kg-1· min-1 of dobutamine
Full size table
Table 3 The response in strain parameters to 10 and 20 μg · kg-1· min-1 of dobutamine on a segmental basis
Full size table
Figure 2
figure 2

Circumferential and radial strain in respect to changes of left ventricular ejection fraction. The figure shows changes in circumferential and radial strain in respect to changes of left ventricular ejection fraction (EF) at rest and with dobutamine stress (10 and 20 μg/kg-1/min-1). Values expressed as mean with standard deviation. LV = left ventricle, EF = ejection fraction.

Full size image

Intra- and inter-observer variability

All parameters were reproducible on an intra- and inter-observer level. Bland Altman plots are displayed in figure 3 and table 4 shows the 95% confidence intervals of the difference between the repeated measurements. Observer variability at rest was best for EccSAX and worst for EllRV as determined by 95% confidence intervals of the difference. Observer variability did not significantly increase with dobutamine stress (table 5).

Figure 3
figure 3

Bland Altman Plots for intra- and inter-observer variability. Bland Altman Plots for intra- and inter-observer variability obtained for all strain parameters at rest on a segmental basis

Full size image
Table 4 Intra- and inter-observer variability of different strain parameters
Full size table
Table 5 Intra- and inter-observer variability of circumferential and radial strain parameters of the LV at rest and with dobutamine stress
Full size table

Comparison with natural radial strain

There was reasonable agreement between mean ErrSAX and natural radial strain as demonstrated in figure 4.

Figure 4
figure 4

Bland Altman Plot showing the relationship between Err SAX and natural radial strain. Bland Altman Plot showing the relationship between ErrSAX (Mean 19.6; 15.8-23.4 95% confidence interval) and natural radial strain (Mean 24; 21.7-26.4 95% confidence interval). Err SAX = left ventricular short-axis radial strain

Full size image

Discussion

The current study includes a unique population of healthy volunteers studied at rest and with DS-CMR and demonstrates several important findings.

Firstly; CMR-FT can quantify wall motion changes between rest and dobutamine stress. Secondly; we noted considerable intra- and inter-observer variability for all parameters; which was most pronounced for RV longitudinal strain and smallest for LV circumferential strain. Thirdly; normal values of CMR-FT cover a large range with considerable overlap between rest and stress between different subjects (Figure 2) and closely correlate with hemodynamic changes secondary to changed LV function.

In essence; our data demonstrate that CMR-FT strain parameters can be derived from routine SFFP cine sequences at varying levels of DS-CMR. The current reference standard for quantitative wall motion assessment with CMR is myocardial tagging [11, 14, 15]. Myocardial tagging based strain assessment has been demonstrated to improve diagnostic accuracy of DS-CMR in patients with suspected CAD as well as in patients with myocardial hibernation [16]. Adding quantitative analysis to DS-CMR in CAD may not only increase diagnostic accuracy as compared to visual analysis for the detection of ischemia during high-dose dobutamine stress [17] but may also detect quantitative changes in myocardial strain already detectable at lower stress levels; thereby increasing feasibility of the test [18].

There is evidence that CMR-FT may also be of clinical utility. Hor et al have recently shown that CMR-FT provides similar results compared to HARP myocardial tagging in a patient population with Duchenne muscular dystrophy [5]. Maret et al demonstrated that CMR-FT can be used in CAD patients to accurately detect strain in the radial and longitudinal direction correlating with the presence of myocardial scarring [4]. CMR-FT is also useful for the assessment of myocardial viability [19]. Detection of quantitative contractile reserve in patients with myocardial hibernation using low dose dobutamine has the potential to predict functional recovery after revascularisation with higher accuracy in the future [20]. In this context; CMR-FT may serve as an additional tool alongside conventional visual analysis; thus facilitating the detection of subtle wall motion abnormalities and the identification of contractile reserve; particularly for the less experienced observer.

The importance of the intra- and interobserver variability documented in the current study needs to be taken into consideration. This has also been reported by echocardiography based speckle tracking studies and our results are similar [21]. The parameter with highest variability at rest in the current study was longitudinal strain of the RV indicating that the analysis of the thin-walled RV with CMR-FT is not yet adequately accurate. This might also be explained by difficulties in endocardial tracking due to difficulty in accurately following the tricuspid valve annulus motion with the CMR-FT software; and RV trabeculations that also lead to greater variability in RV volumetric assessment [22]. The most robust parameter in our study was circumferential strain of the LV; which might be clinically valuable. CMR- FT algorithm allows reliable and easy border tracking; the frame-to-frame displacement of features tracked is equivalent to evaluating the local velocity (ratio between displacement and time interval); allowing automatic evaluation of tissue motion during the cardiac cycle. The tracking results from this algorithm may be more reliable due to the inherently high image quality with CMR. However echocardiographic speckle tracking has better temporal resolution than CMR-FT. In addition our collective consisted of good breath holders and as a consequence we obtained good image quality at an intermediate stress level. It is therefore not surprising that observer-induced variance did not significantly increase with stress. Whether potentially degraded image quality at higher stress levels in patients who are not able to hold their breath would substantially obscure CMR-FT results needs to be prospectively assessed. Interestingly results in radial strain from matching segments from the short-axis and long-axis orientation were not equal. Whether this could be explained by more extensive through-plane motion in the short-axis orientation or increased susceptibility of the 4-chamber view to the breath-holding position of the diaphragm needs to be investigated in healthy volunteers and in patients with scarred areas and segments with wall motion abnormalities. In particular future studies need to investigate whether DS-CMR accuracy could be improved with CMR-FT information that is available with any DS-CMR stress study. There is evidence to suggest that these quantitative parameters have prognostic implications. Stanton and colleagues demonstrated that automated echocardiography speckle-tracking derived global EllLV is a superior predictor of outcome compared to either EF or wall motion score index; and suggested that EllLV may even become the optimal method to assess global left ventricular systolic function [23]. As CMR-FT is a relatively new method such evidence is not yet available and future studies need to investigate whether CMR-FT could also provide such assessment.

Limitations

The sample size of the current study was relatively small. Future studies will need to reassess these parameters in a larger cohort of volunteers and patients. Global Ell; which has been previously shown to be an important; prognostic echocardiographic parameter was only assessed at rest. This was due to time constraints as a whole stack of short axis images had been acquired at each stage of dobutamine for volumetry. Also we did not perform any echocardiographic measurements to compare with CMR-FT data; which needs to be addressed in future studies. Finally the current work aimed to determine the feasibility of CMR-FT during dobutamine stress in a collective of healthy volunteers. Future research needs to prospectively validate this novel technique in pathologies such as coronary artery disease; valvular disease or congenital disorders.

Conclusions

CMR-FT allows derivation of strain mechanics from SSFP cine images at rest and during dobutamine stress CMR corresponding to global hemodynamic changes. The current analysis algorithm requires improvement to reduce observer-induced variance; which at present is comparable to data reported from 2D strain by echocardiography. Within a given CMR lab; this novel CMR-FT technique holds promise for easy and fast quantification of wall mechanics and strain.