Speckle tracking echocardiography-determined measures of global and regional left ventricular function correlate with functional capacity in patients with and without preserved ejection fraction

This study was approved by the University of Florida institutional review board. Adult patients who were referred for an echocardiogram for clinically indicated reasons and who had echocardiographic images that were determined to be adequate for evaluation of regional LV systolic function were approached for enrollment. The patients’ functional status was estimated with the Duke Activity Status Index (DASI) questionnaire [9]. DASI is a 12-item questionnaire which asks if a patient can comfortably complete different activities. Patients receive points for every activity that can be completed comfortably, and points received for each activity are weighted based on the associated metabolic equivalent of each activity. Possible scores range from 0 to 58.2 points. These scores correlate with peak oxygen uptake. Subjects were also classified into groups as follows: controls = no symptoms with physical activity and EF >55% and no evidence of diastolic dysfunction; Group A = no symptoms with physical activity but EF ≤55% or diastolic dysfunction; Group B = symptoms with ordinary activity, such as walking more than 2 blocks; Group C = symptoms with less than ordinary activity. Diastolic dysfunction was defined in accordance with the American Society of Echocardiography recommendations [10]. Patients with moderate or severe valve disease were excluded. Further, those patients with a documented non-cardiac cause of a decreased functional capacity, such as interstitial lung disease, were excluded. In an effort to determine if traditional Doppler echocardiographic parameters of diastolic function correlate better with functional status than do STE measures, a secondary analysis was performed and limited to patients with normal or mildly depressed LV systolic function, defined as an LV EF ≥45%.

Echocardiograms were obtained with an iE33 (Philips, The Netherlands). Parasternal short axis views were obtained at the basal, midventricular, and apical levels [11]. Additionally, three standard apical views (4-chamber, 2-chamber, and 3-chamber) and a parasternal long axis image were obtained. To optimize speckle tracking, images were acquired at as high a frame rate as possible (50–90 frames/s). LV end-systolic and end-diastolic volumes and ejection fraction were determined by manual tracing of the end-systolic and end-diastolic endocardial borders using apical 4- and 2-chamber views, employing Simpson’s biplane method. Pulsed wave Doppler at the level of the mitral valve leaflet tips was used to determine peak early (E) and atrial (A) filling velocities, and deceleration time (DT). Tissue Doppler was used to determine peak early (e’) velocity of the medial mitral annulus. Function of each visualized LV segment was qualitatively evaluated, and a wall motion score was assigned for each LV segment as follows: normal function = 1, mild hypokinesis = 1.5, moderate hypokinesis = 2, severe hypokinesis = 2.5, akinesis = 3, dyskinesis = 4. A wall motion score index was calculated as the average wall motion score for all of the visualized LV segments.

Analysis was performed using QLAB advanced quantification software version 7.1 (Philips, The Netherlands). Automated tracking of myocardial speckles was reviewed and manually adjusted as minimally as possible. The tracking quality of each segment was visually evaluated and if tracking was felt to be inaccurate, strain analysis of that segment was not included. Speckle tracking analysis provided measures of peak circumferential, radial, and longitudinal strain in 18 LV segments (6 basal, 6 mid, and 6 apical) and rotation in the basal and apical LV regions. Strain was determined for the entire myocardial segment, not just the endocardial or epicardial portion. LV twist was calculated as the maximal apical rotation minus the maximal basal rotation. Average global circumferential, radial, and longitudinal strain were calculated. Further, average circumferential, radial, and longitudinal strain in each of the 6 walls of the LV was calculated. If fewer than 2/3 of the variables for a computed variable were available, then that computed variable was not included in the analysis. QLAB software also used the automated endocardial borders to determine the fractional area change (FAC) in each of the LV regions (basal, mid, and apical) by comparing the LV luminal area at end-diastole and end-systole.

Analysis was completed with SPSS Version 20. Continuous variables were compared between groups using a Kruskal Wallis test. Continuous variables were compared between groups A and B using a Mann–Whitney U test. Correlations between continuous variables are presented as Pearson correlation coefficients. The magnitude of correlation coefficients between EF/DASI and EF/regional measures of LV strain were then compared [12]. Categorical variables were compared using Chi-Square or Fisher’s Exact test as appropriate. Multiple linear regression was used to determine if the association between strain variables and the DASI score of functional capacity persisted after adjusting for the baseline variables that were different between the groups (age and glomerular filtration rate). A P value <0.05 was considered statistically significant.

A total of 88 patients were enrolled in this study. Seven patients with moderate or severe valve disease were excluded from analysis. Further, 13 patients with clear non-cardiac causes of a decreased functional capacity, such as interstitial lung disease, were excluded from the analysis. Of the remaining 68 patients, 19 were identified as controls as they had no symptoms and normal LV systolic and diastolic function as described above. The baseline characteristics for the 19 controls and 49 patients with a reduced functional capacity or asymptomatic LV systolic or diastolic function are shown in Table 1. In the 49 patients, the primary indication for the echocardiogram was a history of cardiomyopathy, shortness of breath, or edema in 55%, chest pain in 18%, syncope in 8% and abnormal rhythm in 6%. In the 35 patients with an LV EF ≤55%, 23% had a clear ischemic mechanism for LV systolic dysfunction, whereas 77% did not have a clearly documented ischemic mechanism.

Speckle tracking echocardiography strain analysis was attempted on 1224 LV segments (18 segments for each of 68 patients). Importantly, because of inadequate tracking quality, STE-derived measures of LV function could not be determined for all segments. Circumferential strain could be determined in 913 (75%), radial strain in 781 (64%), and longitudinal strain in 804 (66%) of the LV segments analyzed. To test the inter-observer reliability between the two investigators who completed strain analysis for this project (JP,TN), each independently determined circumferential, radial, and longitudinal strain in a total of 48 LV segments from 8 different subjects. Inter-observer reliability was very strong for longitudinal strain (r = 0.91, average difference 0.5 ± 3%) and good for circumferential strain (r = 0.8, average difference 0.65% ± 4.8%) and was similar to previously reported inter-observer reliability of speckle software [13]. In our study, inter-observer reliability was weakest for radial strain (r = 0.85, average difference 0.17% ± 10%).

In the 49 patients with a reduced functional capacity or asymptomatic LV systolic or diastolic function (Groups A, B, and C), global longitudinal, radial, and circumferential strain had a strong linear association with the DASI score of functional status (Table 2). The linear association between these global measures of strain and the DASI score of functional status remained significant after controlling for age and glomerular filtration rate in a multiple linear regression model, whereas age and GFR did not provide prediction of DASI score independent of the effects of global measures of strain. However, the strength of the correlations between these global strain measures and DASI score was not significantly different than the strength of the correlation of EF with DASI score in this small cohort (Table 2). The strength of the correlations between automated measures of FAC and the DASI score was not significantly different than the strength of the correlation of manually determined biplane EF with DASI score (basal FAC: r = 0.44, n = 44, P = 0.003; mid FAC: r = 0.46, n = 49, P = 0.001; apical FAC: r = 0.44, n = 49, P = 0.002). Similarly, the strength of the correlation of the wall motion score index, determined by qualitative assessment of myocardial contractility in all visualized LV segments, with the DASI score (r = −0.5, P < 0.001) was not different than the strength of the correlation between other global measures of LV function and DASI score. LV twist had a weak correlation with DASI score.

In the patients included in this study, longitudinal, circumferential, and radial strain in the various walls of the LV had significant linear associations with the DASI score of functional status (Table 3). As detailed in Table 3, the linear association between most of these regional measures of strain and the DASI score of functional capacity remained significant after controlling for age and glomerular filtration rate in a multiple linear regression model, whereas age and GFR did not provide prediction of DASI score independent of the effects of regional measures of strain. In nearly all of the walls of the LV, longitudinal strain provided the strongest correlation with functional status. Longitudinal strain in the inferolateral segments had the strongest correlation with the DASI score of functional capacity (r = −0.72, P < 0.001) (Figure 1). The strength of the correlations between most of these regional measures of LV strain and DASI score was not found to be significantly different than the strength of the correlation of EF with DASI score (Table 3). However, the strength of the correlation between anteroseptal longitudinal strain and the DASI score was stronger than the correlation of EF with DASI score in the same subjects, with borderline statistical significance (r = 0.66 vs. r = 0.43, n = 34, P = 0.052).

As shown in Table 3, nearly all STE measures in the various walls of the LV were significantly different between those patients in groups A, B, and C. Further, some regional measures of longitudinal and radial strain were different between those patients without symptoms (group A) and those patients with mild symptoms (group B), whereas EF was not different between these groups (Figure 2). Only one patient without symptoms (group A) had qualitatively determined wall motion abnormalities (global mild hypokinesis). The predominant qualitatively determined wall motion abnormality in those patients with symptoms (groups B, C) was in the anteroseptal, inferolateral, inferior, or inferoseptal walls of the LV. The pattern of qualitatively determined wall motion abnormalities was similar between those patients with mild (group B) and moderate (group C) symptoms.

In the 34 patients included in this study who had an EF ≥45%, biplane EF did not have a linear association with the DASI score of functional status (r = 0.064, P = 0.72, n = 34). Further, traditional echocardiographic measures of “diastolic function”, such as E/e’, had no correlation with the DASI score of functional status (r = −0.16, P = 0.37, n = 32). However, similar to what was shown in all patients, longitudinal strain in the inferolateral segments had significant correlation with DASI score in these patients with preserved EF (r = −0.53, P = 0.03, n = 16).