In this retrospective cohort study of 62 patients with various cardiac conditions we show that LV GPSs on ss-CS cine images were significantly lower than these on segmented cine images. Compared to normal and thickened myocardium, poor correlation of GPSs between two cines was found in thinned myocardium. Both increasing of EF and HR independently augmented absolute underestimation of GPSs respectively. However, relative underestimation of GPSs except GRS only increased with HR increasing. For patients with segmented cine images, GPSsegmented could be calculated by combination of heart rate and GPScs.
Difference of LV volumetric parameters between segmented cine and ss-CS cine
Several studies have already demonstrated that ss-CS cine with breath-holding or free-breathing could acquire high quality images and LV volume assessments were approximately consistent with those from segmented cine in volunteers and patients (4, 9, 14). In our study, the strong agreement and high reproducibility of LV volumetric parameters between the two cines confirmed that ss-CS cine imaging is a reliable cine imaging to evaluate LV volumes in real clinical practice involving different cardiac functional states and different LVWTs.
Difference of LV GPSs between segmented cine and ss-CS cine
Previous study demonstrated that LV GRS, GCS and GLS obtained with ss-CS cine were underestimated in healthy volunteers compared to segmented cine, and relative differences were approximately 28.77%, 16.20% and 30.16%, respectively (8). In this study, although LV GPSs from segmented cine and ss-CS cine in patients were lower than these volunteers, relative differences between both cines were analogous (GRS: 26.55%, GCS: 14.46%, GLS: 27.10%, respectively). According to present study, relative difference of GCS was less than that of GRS and GLS, which is consistent with previous study (8). Contrary to our study, Kido et al showed there was no significant difference in LV GRS between CS cine using 80 iterations and conventional cine (15). In our study, 40 iterations were applied. Although the temporal resolution is very important for strain analysis, the differences of segmented cine (39 ms) and ss-CS cine (46 ms) in our study is not significant (16). So, number of iterations may be the reason of different conclusions. So, CS cine technique should be standardized in clinical practice in future.
Influence of EF, wall thickness and heart rate on underestimation of GPSs from ss-CS cines
Stokke T.M. et al had confirmed that LV EF was highly dependent on longitudinal and circumferential shortening by use of a specifically mathematical model, which means there exists a closely positive correlation between global strains and EF (17). In our study, patients with higher EF had higher absolute differences but similar relative differences of LV GPSs between segmented and ss-CS cine CMR. This indicated that the GPSs from ss-CS cine were proportional to those from segmented cine in all patients irrespectively with EF status.
A previous study showed severe aortic stenosis patients with different wall thickness had different GLS, which implied wall thickness is also related to myocardial strain (18). Rodrigues J.C. et al demonstrated that increasing LVWT was accompanied by reduced myocardial longitudinal and circumferential shortening measured by tagging CMR in hypertensive heart disease (19). Our study demonstrated that absolute underestimation in GRS and GCS were different among different degrees of wall thickness. We also found that the correlation of GRS in thinned myocardium were lower than those in normal and thickened myocardium. This finding might be because GCS and GLS is measured with less couture correction, but erroneous definition of endocardium and epicardium has great impact on GRS (20).
Boettler. P. et al. showed GPSs in children was strongly correlated with HR. GPS increased significantly with decreasing HR (21). Our findings showed that nearly all absolute and relative differences in GPSs between both sequences increased as HR increased, which is consistent with previous real time cine technique (8, 22). This may be explained by that ss-CS cine images acquired less raw data with increased HR. High HRs significantly shortened cardiac phase causing a degradation in image quality (23) and the strain values may be underestimated with lower image quality (24).
Patient-related factors to independently predict underestimation of GPSs from ss-CS cine
In our study, absolute underestimations of GPSs are both influenced by EF and HR. Relative underestimations of GPSs could be easily adjusted only according to HR. Other patient-related factors including sex, age, height, weight, body mass index, body surface area, LVWT and EF had no contribution to the relative underestimation of GPSs, which is consistent to ss-CS cine in volunteers and other real time techniques (8, 22). From this point, relative adjustment is preferred during LV strain analysis obtained from ss-CS cine.
Limitation
Our study has several limitations. First, image quality of ss-CS cines is different with increasing times of iteration reconstructions (25), and times of iteration reconstructions were 40 in this study. The influence of number of iterations on the accuracy of LV strain need to be investigated. Second, GPS values are significantly different among various post-processing software, and not changeable between different software applications. Therefore, popularization of our findings to other software need to be validated. Moreover, underestimation of myocardial strain by CS cine CMR heavily depends on CS reconstruction methods as well as CS acquisition parameters. The value of detailed analyses of patient-related factors using the current acquisition parameters and reconstruction method may rapidly diminish with the refinements of CS image acquisition and reconstruction. But it is still very important to the readers who apply similar cine CS technique as ours.