This is the first study that uses TPM to assess species differences in regional myocardial function. Importantly, we here provide the first detailed assessment of normal segmental cardiac function in two small animal models, commonly used in cardiovascular research, and compare them to humans. Contrary to the fundamental assumption of similar regional myocardial function across the various animal models, our findings demonstrate that there are significant species differences already under physiological baseline conditions. Therefore, genetical or surgical interventions typically used to investigate pathological cardiac function in diseases and their underlying mechanisms could lead to different outcomes in animal models as compared to humans.
Specifically, our main findings are:
The peak radial velocities are fairly homogeneously distributed throughout the left ventricle in mice and humans, while they are significantly higher in base and mid-ventricle in rabbits.
The base is moving towards the apex in all species, but peak-velocities are highest in inferior / septal segments in mice, and lateral segments in humans, and homogeneously in basal segments in rabbits, respectively.
Rotational velocities in mouse hearts are positive at early systole (indicating clockwise rotation when viewed from apex to base) but negative in rabbits and humans (i.e. counter-clockwise rotation).
The torsion (i.e. the net twist) between base and apex is similar across all species. However, the temporal evolution and relative contributions between base and apex varies within species.
Importantly, we are first to demonstrate significant differences between mice and humans in the early systolic rotational motion of the LV and to provide a comprehensive comparison of segmental motion. Two tagging studies were published so far comparing myocardial motion across species. Henson et al. reported on the LV twist in mice and humans
 with a low temporal resolution (9.4 and 42 ms for mice and humans) and a low spatial tag resolution (1.2 and 7 mm for mice and humans) only demonstrating the same rotational behavior as a sum over the systole. Furthermore, only ventricular twist and strain were quantified without performing a segmental analysis. Liu et al. performed a comparison between mice, rats, and humans
 but with a relatively low temporal resolution (8, 14, and 37 ms for mice, rats, and humans) compared to the time scale of myocardial motion, and with low spatial tag resolution (0.6, 0.9, and 6 mm for mice, rats, and humans, respectively). Moreover, only ventricular twist and strain were quantified without a segmental analysis. Both tagging studies
[32, 33] confirmed the same opposite rotation of apex and base during mid systole as we report across species, e.g. with a systolic clockwise net twist of the base and a counter-clockwise twist of the apex when viewed from apex to base. The net twist angle between base and apex in our study was similar in mice, rabbits and humans (~8°, with somewhat higher standard deviations in mice) as recently reported by Zhou et al.
As mentioned above, the global LV counter-clockwise rotation during early systole in humans and rabbits contrasting with a global clockwise rotation in mice observed in our study, however, has not yet been reported to date. This is likely due to the fact that in the mice studies from Henson and Liu A) the tagging module (7–10 ms) was played out after the detection of the R-wave and B) the temporal resolution (8–9.4 ms) was much lower than in our study (4.6 ms; with the acquisition of the first cardiac frame 2.1 ms (TE) after the R-wave) thereby causing the absence of the first three cardiac frames that demonstrate this different rotational behaviour during early systole in our mice measurements. Furthermore, the lower spatial resolution provided by the tag grid pattern may limit the detection of these subtle short-range differences in the rotational motion pattern
[32, 33]. Despite the same limitations (the tagging module after the R-wave and the low temporal resolution of 8–10 ms), a similar twist behaviour in all slices (as shown in Figure
6) can also be observed in the data by Li et al.
More significant regional differences between species were detected that have not been reported so far. Most pronounced segmental differences occurred for the long-axis velocities as evidenced by Figure
5. Noticeably, the lateral-septal difference for the systolic long-axis velocities was not visible in rabbits and was even reversed in mice compared to humans (see Figure
5A). Less pronounced differences between species were seen when comparing anterior with inferior segments (see Figure
5B). Reduced murine radial and long-axis peak velocities could be found for more apical regions in rabbits and humans compared to basal segments as corroborated by the analysis of basal-to-apical velocity ratio in Figure
5C. The comparison between the ratios of different myocardial segments was chosen due to its dimensionless nature, which therefore does not depend on absolute velocity values. The results demonstrate the importance of regional cardiac function, which revealed subtle but significant differences within species. Furthermore, the different rotational behaviour results in a different evolution of twist angles during the cardiac cycle in all three slices in mice compared to humans and rabbits as can be seen in Figure
Our findings have significant implications.
Particularly, the observation of a different twist and regionally inhomogeneous motion patterns across species is very important for investigating (transgenic) models of cardiac diseases, where a different twist may lead to different strain patterns and potentially a different remodelling. Our results also show a higher wall thickness to radius ratio in mice, which is potentially relevant for understanding the different rotational patterns across species as speculated by Liu et al. . This is also in line with the findings presented in Figure
7, which illustrates that the endo-to-epi twist ratio is larger in mice, than in the other two species.
Pronounced species differences have been reported in cardiac ion channel function, action potential shape
[13, 14], Ca2+ cycling proteins, and electromechanical coupling
 between rodents and humans. However, it is not clear how these differences impact on myocardial contractile function. Although this question cannot be answered by this study, our results clearly demonstrated that significant differences exist. Rabbits, which have cardiac ion channels
 and Ca2+ cycling properties
 that are more similar to humans than mice also show a closer similarity in myocardial motion patterns. Consequently, rabbit models might be more suitable to mimic the human phenotype of diseases with electrical and mechanical impairment than mice
. Indeed, several transgenic rabbit models of human diseases such as long-QT syndrome and hypertrophic cardiomyopathy have demonstrated pronounced similarities with the human phenotype
[5, 6]. It might be of further interest to investigate other mammalian hearts such as guinea pigs, rats, pigs or dogs that are also commonly used for the exploration of cardiovascular diseases to complete our understanding of species differences in myocardial function.
Besides molecular differences in ion channels and calcium cycling and contractile proteins, structural differences such as the myofiber orientation may contribute to species differences in myocardial function. Streeter et al. described the ventricular myocardium as a continuum in which myofiber orientation varied smoothly across the ventricular wall with a fiber angle variation of up to 180° transmurally
. Measurements of myofiber orientation throughout the walls of the RV and LV in different mammalian hearts such as dogs
, and humans
 have shown that there is significant local variation of fiber orientation, particularly at the junctions of the RV and LV free walls and in the interventricular septum, which might explain the regional differences and the different rotational characteristics in our study. Recently, Healy et al. published a study comparing the myofiber structure between mice, rabbit, and sheep using diffusion tensor CMR
. Most pronounced (and significant) differences were found between mice and both other species. Notably, an opposite helix angle at the anterior junction of the RV and LV was observed in mice.
One can assume that each excitation should give rise to bi-phasic twisting. However, the missing initial un-twisting in mice might be due to the thin ventricular wall (~1.5 mm) and the short conduction delay between endo- and epicardium (~2 ms)
 not allowing a staged mechanical activation of initially endo- and later epicardial fibers. If the entire LV wall is activated, epicardial fibers dominate (lager mass fraction) resulting in the observed rotational behavior. This may be further established e.g. by a high-speed video of an isolated beating mouse heart.
However, in order to create a link between fiber structure and myocardial contraction, exhaustive computational models of electrical and mechanical function are necessary
. A detailed comparison of such differences with respect to the myocardial motion pattern is warranted to thoroughly explain the observed species differences in myocardial motion patterns.
One limitation is given by the fact that a direct comparison between different species is difficult to perform due to differences in absolute myocardial velocities. Moreover, only the myocardial motion pattern with its velocity amplitudes (or velocity ratios) was investigated. Due to the varying heart rates the velocity time courses were normalized to end-systole in order to maintain characteristic features of the motion pattern, particularly during diastole. To allow a comparison of relative temporal differences (of time-to-peak values), systole and diastole should be normalized to account for different heart rates. Therefore, a retrospectively ECG-triggered sequence would be beneficial to reduce errors due to the normalization of the cardiac cycle. However, the analysis of temporal parameters was beyond the scope of this study.
Anesthesia was necessary for CMR experiments in mice and rabbits but was not used in human subjects. Since anesthesia is known to affect blood pressure, heart rate, and cardiac contractility
[40, 41], the necessity of anesthesia in animal studies may impact on some aspects of regional and global myocardial mechanical function. However, a recent study systematically assessing hemodynamic effects of various anesthetic drugs in mice and rats demonstrated similar global systolic and diastolic myocardial function under isoflurane or ketamine/xylazine anesthesia as assessed by peak rate of pressure rise (dP/dtmax) and decline (dP/dtmin), stroke work and relaxation time - despite differential effects on blood pressure and heart rate
. It is expected that the pronounced species differences in regional myocardial mechanical function between mice and rabbits – particularly those in rotational behavior – are thus not due to different anesthetic regimen; however, the effect of anesthetics cannot be definitely ruled out.