Myocardial T2 mapping reveals age- and sex-related differences in volunteers

The ethical board of Heinrich Heine University Düsseldorf approved the present study
(application number 4307). All participants were enrolled after informed consent was
obtained. The study complies with the declaration of Helsinki.

Volunteers

In total 74 volunteers were enrolled for the study. Depending on their age, two groups
were formed (35 years; younger volunteers, 35 years; older volunteers) and compared
in Table 1. Volunteers were characterized by lacking of any cardiovascular disease, no symptoms
of inflammation, prohibition of alcohol intake 48 h, absence of hereditary cardiovascular
diseases, and normal electrocardiogram. Additionally, older volunteers without a medical
history of coronary artery disease were seen in our regular ambulance. They received
regular functional CMR including stress testing and Late Gadolinium Enhancement (LGE).
In the present study we included volunteers without structural myocardial disease
as delineated by LGE and without perfusion defects indicating a relevant coronary
artery disease as identified by perfusion imaging and without atrial fibrillation.
CMR findings were analysed in a blinded fashion by two experienced observers. The
group of older volunteers showed a typical appearance of cardiovascular risk factors
like hypertension and diabetes (Table 1) which were not defined as exclusion criteria.

Table 1. Volunteer characteristics

CMR

Phantom measurements as well as volunteer scans were performed using a 1.5 T scanner
(Archieva, Philips, Best Netherlands) with a 32-channel phased array coil.

For T2 mapping we applied the GRASE sequence which has already been used for quantitative
T2 measurement in liver and brain 18]-20]. This sequence combines the TSE and echo-planar imaging methods by using a train
of refocusing 180° pulses and an odd number of additional gradient echoes for each
spin echo. This sequence was used with cardiac triggering and respiration navigator
gating with the following parameters: TR?=?1 RR interval, number of echo images?=?15,
echo spacing 10 ms, leading to an echo train of 150 ms, number of gradient echoes
for segmented acquisition?=?3 (EPI factor), FA?=?90°, spatial resolution: 2 × 2 ×
10 mm³, parallel imaging (SENSE) with an acceleration factor of 2, k-space data acquired
with cartesian encoding scheme. For blood saturation a double inversion (black-blood)
pulse was used.

Phantom experiments

Detailed information about the phantom experiments can be found in the Additional
file 1. We validated GRASE-derived T2 values with a conventional multi echo spin echo (MESE)
sequence in commercially available cow meat (Additional file 1: Figure S1) in order to find possible effects of EPI-like segmentation on T2.

Besides validation purposes, the phantom study was set up to test the sensitivity
of myocardial T2 quantification in terms of focal and global tissue water changes.
In the first experiment, cow muscle (M. erector spinae) was cut into 15 small pieces
of 20 g and 14 pieces were dehydrated by vacuum concentration (SpeedVac, Thermo Fisher
Scientific) at given durations (0.5, 1, 2, 3, 4, 5, 6, 7, 8, 10, 11, 15, 20, 24 h).
All pieces of meat were imaged at once with GRASE. The percentage of tissue water
was calculated as follows: the difference of wet weight to dry weight was divided
by wet weight. This ratio (0.63) was constant between 20 and 24 h, indicating complete
tissue drying and the relative tissue water mass of the meat to be 63%. This value
was set to 100% of tissue water.

Volunteer scans

The measurement protocol was conducted in the following order: scout and reference
scans, cine-loops in 3 short axis (SAX) slices and in four chamber view (4CH) (balanced
SSFP cine-MRI: TR/TE?=?2.9/1.5 ms, FA?=?60°, res?=?8 × 1.5 × 1.5 mm³, 35 phases, breath-hold), T2-weighted imaging in the respective SAX (TSE-STIR: TR?=?2
RR-intervals, TE?=?58 ms, spatial resolution?=?2 × 2 × 10 mm³, imaging in mid-diastole, breath-hold), GRASE in the respective 3 SAX (GRASE: TR?=?1
cardiac cycle, EPI factor 3, 15 echos with 10 ms inter echo spacing, FA?=?90°, resolution?=?2
× 2 × 10 mm³, end-diastolic trigger), and strain-encoded imaging in two chamber view (2CH), 4CH
and the respective SAX (SENC imaging: TR?=?25 ms, TE?=?0.9 ms, FA?=?30°, spatial resolution:
2.5 × 2.5 × 10 mm³). Older volunteers (35 years), enrolled in our regular ambulance, additionally received
Gadolinium DTPA (Gadovist, Bayer Healthcare) for myocardial stress-testing and LGE.
The latter was performed using a 3-dimensional gradient -spoiled turbo fast field
echo sequence with an nonselective 180° inversion-recovery pre-pulse-triggered to
end-diastole which covered the whole ventricle (TR/TE?=?3.2/1.16 ms; FA?=?15°; spatial
resolution: 1.5 × 1.7 × 10 mm³, patient-adapted prepulse delay?=?200–300 ms, breath-hold).

T2 mapping post processing

T2 maps of the respective object (phantom or volunteer) were generated off-line using
software based on the graphical programming language LabVIEW (National Instruments,
Austin, TX). An exponential decay curve was fitted to the intensity decline of each
pixel within the images obtained from the multi echo sequence (Figure 1A). The bias was calculated from the noise of all echo images and assumed to be constant,
so that the problem could be linearized and the regression coefficient (R²) could be used as a goodness-of-fit parameter in order to exclude accidental values.
If R² was not within a tolerance interval chosen to be 0.7?1, the corresponding T₂ value was not considered for further calculations. The resulting T₂ constants were colour-coded using the spectral look-up table (Figure 1B).

Figure 1. Principle of T2 map calculation. (A) Example of 6 GRASE-derived echo images with echo times given in the figures. Note,
there is only very little cardiac motion between 10 and 150 ms. (B) An exponential fit was performed for each image pixel (exemplarily for the blue circle
in the anterior wall of A) with amplitude (h) and damping (1/T₂) as fit parameters (fixed bias, y₀). T2 value calculation in myocardial tissue was performed for 5, 10 and 15 echoes
with 10 ms interecho time. T2 maps are illustrated by a colour-coded map, assigning
0 ms to black and 150 ms to red.

Data analysis for T2 quantification and strain encoded imaging

Endo- and epicardial contours were manually drawn in reference images and automatically
overlayed to the T2 map. Segmentation was done off-line using the same software as
for the generation of T2 maps 7]. For this, a starting point was chosen at the posterior insertion of the right ventricle.
Depending on the SAX location, 4 or 6 segments were generated automatically in a clockwise
manner according to the 17 segment model of the AHA omitting segment 17 21]. The contours were drawn by two independent observers omitting “slow flow” ventricular
blood or epicardial fat. The coefficient of variation (CoV) was calculated as the
ratio of the standard deviation of the inter- or intraobserver differences divided
by the mean of the measurement. Time interval between intraobserver analysis was 28?±?5 days.

To evaluate the role of wall movement on T2 values within the same segment, we acquired
myocardial strain values by strain encoded imaging (SENC) of each myocardial segment
analysed. Strains were derived from SENC images using dedicated software (Diagnosoft
MAIN, version 1.06, Diagnosoft Inc., Palo Alto, California). Circumferential strain
was calculated for 12 myocardial segments, not considering segments 2, 5, 8, 11, and
17. Peak systolic strain was defined as minimal strain value of all time frames, while
peak diastolic strain was characterized as minimal strain value of all time frames
following early diastolic relaxation. Early diastolic circumferential strain rate
(E_cc/s) was used to measure early diastolic function and was calculated in 4 midventricular
segments by dividing the change in strain between time frames by the temporal resolution
(25 ms). E_cc/s was defined as the slope from end-systole to mid-diastole as has been previously
published 22].

Statistics

Unless otherwise stated, data are presented as mean value?±?standard deviation (SD).
Data were statistically analysed by the paired or unpaired Student’s t-test. Regression analysis was done with Pearson product moment correlation. P values
below 0.05 were assumed to be significant.