Patients Characteristics: Among 162 patients enrolled for the present study, 11 patients were excluded because adequate speckle tracking was not achieved due to frequent extra beats (6 cases) or very poor image quality (5 cases). The final study population consisted of 151 (93.2%) patients (age: 65 ± 14 years, male gender: 117 patients (77.5%)). The major reason for the echocardiography was coronary artery disease (CAD) in 58 patients (38.4%), arrhythmia in 33 patients (21.9%), valvular heart disease in 32 patients (21.2%) and others in 12 patients (8.0%). Mean LVEF was 60.2 ± 9.6% and 24 patients (15.9%) had LVEF < 50%.
GLS and Ultrasound Transmit Frequency
The mean value of GLS-HRES in the study patients was − 17.9 ± 4.4%, GLS-HGEN was − 18.8 ± 4.5% and GLS-HPEN was − 18.8 ± 4.4% (Table 1). There were good correlations among these GLS values (GLS-HRES vs. GLS-HGEN, r = 0.96 [95% CI, 0.95 to 0.97], p < 0.0001; GLS-HRES vs. GLS-HPEN, r = 0.95 [0.93 to 0.96], p < 0.0001; GLS-HGEN and GLS-HPEN, r = 0.97 [0.96 to 0.98], p < 0.0001) (Fig. 2).
Table 1
Global longitudinal strain and ultrasound transmit frequency.
| N | GLS-HRES | GLS-HGEN | GLS-HPEN | p |
All patients | 151 | -17.9 ± 4.4% | -18.8 ± 4.5%‡ | -18.8 ± 4.4%‡ | < 0.0001 |
Image Quality | | | | | |
Optimal | 104 | -19.0 ± 4.1% | -19.7 ± 4.2%‡ | -19.7 ± 4.2%‡ | < 0.0001 |
Suboptimal | 47 | -15.7 ± 4.4% | -17.0 ± 4.5%‡ | -16.9 ± 4.2%‡ | < 0.0001 |
Etiology | | | | | |
Coronary artery disease | 58 | -17.4 ± 4.3% | -18.5 ± 4.4%‡ | -18.4 ± 4.6%‡ | < 0.0001 |
Others | 93 | -18.3 ± 4.5% | -19.0 ± 4.5%‡ | -19.1 ± 4.3%‡ | < 0.0001 |
LVEF | | | | | |
< 50% | 24 | -12.3 ± 3.4% | -12.8 ± 3.7%* | -12.8 ± 3.4%† | 0.003 |
≥ 50% | 127 | -19.0 ± 3.8% | -20.0 ± 3.6%‡ | -19.9 ± 3.6%‡ | < 0.0001 |
Continuous variables are expressed as mean ± standard deviation.
GLS denotes global longitudinal strain; GLS-HRES, GLS measured by HRES preset; GLS-HGEN, GLS measured by HGEN preset; GLS-HPEN, GLS measured by HPEN preset; LVEF, left ventricular ejection fraction. P value for difference among 3 frequency presets.
* p < 0.05, †p < 0.01, ‡p < 0.0001 vs. GLS-HRES.
Correlation between GLP-HRES and GLS-HGEN (upper left), between GLS-HRES and GLS-HPEN (upper right) and between GLS-HGEN and GLS-HPEN (lower).
GLS denotes global longitudinal strain; GLS-HRES, GLS-HRES, and GLS-HPEN, GLS measured by the HRES, HGEN and HPEN preset, respectively.
There was significant difference in GLS among three frequency presets (p < 0.0001), and the magnitude of GLS-HRES was significantly smaller than GLS-HGEN (p < 0.0001) and GLS-HPEN (p < 0.0001) (Table 1, Fig. 3). No difference was observed between GLS-HGEN and GLS-HPEN (p = 0.61). The absolute and relative differences between GLS-HRES and GLS-HGEN were 1.1 ± 1.0% and 6.9 ± 7.2%, respectively. Those between GLS-HRES and GLS-HPEN were 1.1 ± 1.2% and 7.1 ± 8.6%, respectively (Table 2). Bland-Altman plots showed that there was no systematic difference between any pairs among 3 presets (GLS-HRES and GLS-HGEN, p = 0.57; GLS-HRES and GLS-HPEN, p = 0.82; GLS-HGEN and GLS-HPEN, p = 0.34) (Fig. 4).
Box-and-whisker plots of global longitudinal strain (GLS) by 3 frequency presets. GLS-HRES, GLS-HRES, and GLS-HPEN denotes GLS measured by the HRES, HGEN and HPEN preset, respectively.
Table 2
Absolute and relative difference in GLS among HRES and other presets
GLS-HRES vs. GLS-HGEN |
| Absolute difference | p | Relative difference | p |
All patients | 1.1 ± 1.0% | | 6.9 ± 7.2% | |
Image Quality | | | | |
Optimal | 1.0 ± 0.7% | 0.004 | 5.4 ± 4.0% | < 0.0001 |
Suboptimal | 1.5 ± 1.4% | 10.3 ± 10.7% |
Etiology | | | | |
Coronary artery disease | 1.3 ± 1.0% | 0.20 | 7.6 ± 6.4% | 0.33 |
Others | 1.1 ± 1.0% | 6.5 ± 7.6% |
LVEF | | | | |
< 50% | 0.8 ± 0.6% | 0.10 | 7.0 ± 4.4% | 0.97 |
≥ 50% | 1.2 ± 1.0% | 6.9 ± 7.6% |
GLS-HRES vs. GLS-HPEN |
| Absolute difference | p | Relative difference | p |
All patients | 1.1 ± 1.2% | | 7.1 ± 8.6% | |
Image Quality | | | | |
Optimal | 1.0 ± 0.9% | 0.11 | 5.7 ± 4.8% | 0.002 |
Suboptimal | 1.4 ± 1.6% | 10.2 ± 13.3% |
Etiology | | | | |
Coronary artery disease | 1.1 ± 1.0% | 0.86 | 6.4 ± 5.6% | 0.48 |
Others | 1.2 ± 1.3% | 7.5 ± 10.1% |
LVEF | | | | |
< 50% | 0.7 ± 0.7% | 0.04 | 6.1 ± 6.3% | 0.54 |
≥ 50% | 1.2 ± 1.2% | 7.3 ± 9.0% |
Continuous variables are expressed as mean ± standard deviation. Relative difference was calculated as [absolute difference / GLS-HRES] x 100 (%). P value for the difference between subgroups. GLS denotes global longitudinal strain; GLS-HRES, GLS-HGEN, GLS-HPEN, GLS measured by HRES-, HGEN- and HPEN preset, respectively; LVEF, left ventricular ejection fraction.
The bias is shown by the solid lines and the levels of agreement (1.95 x SD) by dotted lines.
GLS and Image Quality. Image quality was suboptimal in 47 patients (31.1%). Significant difference in GLS among 3 presets was observed both in patients with optimal quality (p < 0.0001) and in those with suboptimal image quality (p < 0.0001) (Table 1, Fig. 5). GLS-HRES was significantly lower than GLS-HGEN and GLS-HPEN in both groups (Table 1). A significant interaction was observed between GLS variability by frequency and image quality (p = 0.013 by mixed design ANOVA). The absolute and relative difference between GLS-HRES and GLS-HGEN (1.5 ± 1.4% and 10.3 ± 10.7%, respectively) was larger in patients with suboptimal image quality had larger than in those with optimal quality (1.0 ± 0.7%, p = 0.004, and 5.4 ± 4.0%, p < 0.0001). No difference was observed in absolute difference between GLS-HRES and GLS-HPEN between 2 subgroups (1.4 ± 1.6% vs. 1.0 ± 0.9%, p = 0.11), whereas the relative difference was significantly larger in patients with suboptimal image quality (10.2 ± 13.3% vs. 5.7 ± 4.8%, p = 0.002) (Table 2).
Box-and-whisker plots of GLS-HRES, GLS-HRES, and GLS-HPEN in 104 patients with optimal image quality (left) and in 47 patients with suboptimal quality (right). Difference in GLS among presets was observed in both subsets.
GLS and Ischemic Etiology or Ejection Fraction. A significant difference in GLS among 3 presets was observed both in 58 patients with CAD (p < 0.0001) and in others (p < 0.001), and the magnitude of GLS-HRES was smaller than that of GLS-HGEN and GLS-HPEN in both groups (Table 1). No difference in LVEF was observed between patients with CAD and others (CAD, 59 ± 10%; others 61 ± 9%, p = 0.28). A significant difference in GLS was observed among 3 presets in 24 patients (15.9%) with impaired LVEF (p = 0.003) and in 127 patients with preserved LVEF (p < 0.0001). The magnitude of GLS-HRES was smaller than that of GLS-HGEN and GLS-HPEN in both groups (Table 1). There was no interaction between variation of GLS among presets and CAD category (p = 0.35 by mixed design ANOVA) or LVEF category (p = 0.24). No difference was observed in the absolute and relative differences in GLS among these subgroups, except the absolute difference between GLS-HGEN and GLS-HPEN in LVEF subgroups (Table 2).