The initial study group included 102 patients. 18(18%) patients were excluded due to atrial fibrillation in 5(5%) patients, and significant mitral valve disease in 13(13%) patients [6 patients with significant rheumatic mitral stenosis, and 7 patients with moderate to severe mitral regurgitation due to leaflet tethering], Accordingly, the final study cohort consisted of 84 heart failure patients [age: 57.6±8 years, 28(33%) females, NYHA: 2.3±0.6, EF: 55±15%]. Of these, 52(62%) patients had HFpEF (EF: 65±5%), and 32(38%) patients had HFrEF (EF: 39±9%). In addition, 42 young normal subjects [age: 27±4 years, 5(12%) females, EF: 66±5%], and 26 age matched controls [age: 53±12 years, 28(65%) females EF: 61±4.7%] were included. Table 1 shows comparisons between all groups regarding baseline demographic clinical and echocardiographic variables. Briefly, patients with HFpEF were the oldest and had more risk factors, while patients with HFrEF had worse systolic and diastolic echocardiographic parameters.
Pre and Post-Ejection phases in patients and controls:
Table 1 summarizes comparisons between patients and controls for septal, lateral, and averaged mitral annular velocities including pre-ejection (PRE) and post-ejection (POE) variables. It was found that, patients with heart failure had lower PREp, and POEn, higher PREn and POEp, and lower ratios PRE-P/N and POE-N/P compared to controls. Among heart failure patients, HFrEF tended to have lower PREp, and POEn compared to HFpEF (PREp: 6.3±2.2 vs. 4.3±1.6, p<0.001; POEn: 3.3±1 vs. 2.4±0.7, all p<0.001) while there was no significant difference between PREn and POEp compared to HFpEF (PREn: 3.7±1.1 vs. 3.3±1.4, p=0.225 ; POEp: 3.3±1 vs. 2.4±0.7, p=0.248) and, similarly, there was no significant difference regarding both ratios (POE-P/N: 1.8±0.7 vs. 1.5±1.03, p=0.282; POE-N/P: 1.3±0.5 vs. 1.1±1.04, p=0.203). On the other hand, with the exception of a higher POEp in the normal subjects, none of the PRE or POE velocities were different between normal subjects and matched controls. Moreover, it was noted that normal and age matched controls had almost always a predominantly positive PRE and predominantly negative POE, while predominantly negative PRE and predominantly positive POE occurred almost always in HF patients (Figure 2)
Concordance between biphasic velocities of pre and post-ejection phases in patients and controls:
Next, all patients and controls were compared based on our definition of concordance between both biphasic velocities (Normal: PRE predominantly positive and POE predominantly negative, reversed PRE: both PRE and POE predominantly negative, reversed POE: both PRE and POE predominantly positive, and complete reversal: PRE predominantly negative and POE predominantly positive, Figure 2). It was found that almost all normal subjects and matched controls exhibited normal concordance [normal subjects: 39 (93%) subjects, and matched controls: 23(96%) patients], while in normal subjects, 3(7%) subjects exhibited abnormal concordance (reversed PRE in 1 reversed POE in 1, and total reversion in 1), and in matched controls one (4%) patient exhibited reversed POE.
Importantly, abnormal concordance was found in significantly more HF patients compared to controls [49(58%) patients vs. 4(6%) controls, p<0.001]. Abnormal concordance occurred in 24(46%) HFpEF patients and 25(67%) HFrEF patients, (p=0.003). All types of concordance were observed in patients with HFpEF, however, in patients with HFpEF, almost all patients had abnormal concordance [12(44%) reversed POE, and 13(48%) totally reversed concordance]. Importantly, concordance seemed to be significantly associated with NYHA functional class such that the best NYHA was in normal concordance (1.6±1) and the worst was in reversed concordance (2.6±0.5), while reversed PRE and POE was in between normal and reversed (2±1.2 and 2.3±0.6, respectively). Similar patterns were observed in parameters of diastolic and systolic functions where the highest E/A and E/e’, the lowest average e’, the largest ESV and EDV and the lowest EF were noted in reversed concordance, while all these parameters were noted to be best in normal concordance and were intermediate between both groups in reversed PRE and POE. Importantly, in heart failure patients, bundle branch block was not significantly different between groups, while intraventricular conduction delay was most frequent in patients with totally reversed concordance (table 2). Moreover, electrocardiographic QRS duration was lowest in patients with normal concordance, larger in patients with reversed concordance among which it was largest in patients with totally reversed concordance (table 2).
Correlations for biphasic pre and post-ejection waves:
As suggested by the concordance of the biphasic waves, the main waves in the normal pattern were the positively directed in PRE and negatively directed in POE. Correlations noted in our study between different PRE and POE velocities are summarized in table 3 and Figure 3. When correlation was checked between PREp and POEn it was found that both waves correlated significantly in all study subgroups, while the correlations were stronger in normal and matched controls (0.52, 0.79, respectively, all p<0.001) and HFpEF (r=0.56, p<0.001), it was slightly weaker in patients with HFrEF (r=0.42, p=0.018). On the other hand, correlations between PREn and POEp were mainly noted in heart failure patients as no correlation was noted in normal subjects (r=0.25, p=0.107), and only a weak correlation was noted in matched controls (r=0.42, p=0.042). The correlation became slightly stronger in HFpEF (0.43, p=0.005) and was strongest in HFrEF (r=0.611, p<0.001). As such it was noted that the pattern of correlation was opposite in controls and heart failure between signals associated with normal concordance and signals associated with reversed concordance showing the normal concordance signals correlate better in controls and reversed concordance signals correlate better in HF patients.
Correlation between the ratios PRE-P/N and POE-N/P, however, showed significant strong correlations in normal young subjects, weaker correlations in matched controls and HFpEF, while it was the strongest in HFrEF.
Correlations of the pre and post-ejection waves versus E/e’:
Correlations noted in our study against left ventricular filling pressures as expressed by E/e’ are summarized in table 3 and Figure 3. Importantly, it was shown that, most of the correlations of the biphasic velocities were noted in HF patients and not in controls. Briefly negative correlations were found between E/e’ and PREp in patients with HFpEF and HFrEF (r= -0.51, -0.4, p<0.001, =0.025), between E/e’ and POEn in patients with HFpEF only (r= -0.43 , p=0.002), between E/e’ and POEp in patients HFrEF only (r= -0.36, p=0.045), while PREn did not correlate with E/e’ in any of the patients. The ratio PRE-P/N was found to correlate the strongest with E/e’ in both HFpEF and HFrEF (r= -0.55, -0.54; p<0.001, =0.002, respectively), while the ratio POE-N/P showed similar however weaker pattern of correlations (r= -0.3, -0.45, p=0.034, 0.012, respectively).
Receiver operator characteristic curves (figure 4) showed that the best predictors for E/e’>13 was PRE-P/N ≤1.68 (AUC:0.782, sensitivity: 86%, specificity 50%) while that for POE-N/P was ≤1.15 (AUC:0.791, sensitivity: 82%, specificity 64%)
Reproducibility parameters are summarized in table 4. For inter-observer variability, the absolute differences and interclass correlation coefficient for PREp were 0.03±0.93 cm/s, 0.94, respectively, for PREn were 0.72±0.67 cm/s, 0.84, respectively, for POEn were -0.53±0.38 cm/s, 0.91, respectively, and for POEp were 0.01±0.38 cm/s, 0.88, respectively. On the other hand, for intra-observer variability, the absolute differences and interclass correlation coefficient for PREp were 0.11±0.57cm/s, 0.97, respectively, for PREn were -0.22±0.45 cm/s, 0.93, respectively, for POEn were -0.17±0.31cm/s, 0.93, respectively, and for POEp were 0.02±0.34 cm/s, 0.89, respectively.