Interchangeability of transthoracic and transesophageal echocardiographic right heart measurements in the perioperative setting and correlation with hemodynamic parameters

Reduction of right ventricular (RV) function after cardiac surgery has been shown to impact outcomes. Conventional indices for right ventricular dysfunction are validated using transthoracic echocardiogram (TTE) which has limited use compared to transesophageal echocardiogram (TEE) in the perioperative settings. The aim of this study was to assess the agreement of RV systolic function assessment with TEE compared to TTE and assess the association of echocardiographic parameter with hemodynamic indices of RV dysfunction. This was a single center prospective observational study in an academic institution. Fifty adult patients undergoing elective cardiac surgery were included. TTE, TEE and stroke volume measurements pre-cardiopulmonary bypass (CPB) and post-CPB were performed. The variables of interest were anatomical M-mode tricuspid annular plane systolic excursion (AMM-TAPSE), fractional area change (FAC), tricuspid annular velocity (S’) and myocardial performance index (MPI). FAC and AMM-TAPSE measured at the mid-esophageal 4 chamber view had substantial agreement with the TTE acquired parameters (Lin’s concordance correlation coefficient (CCC) = 0.76, 95%CI 0.59–0.86 and CCC = 0.85, 95%CI 0.76–0.91). S’ was significantly underestimated by TEE (CCC = 0.07, 95%CI 0.04–0.19) and MPI showed moderate agreement (CCC = 0.45 95%CI 0.19–0.65). Despite the significant changes in echocardiographic parameters, there were no corresponding changes in stroke volume (SV) or pulmonary artery pulsatility index at the post-CPB period. TEE acquired FAC and AMM-TAPSE had substantial agreement with pre-operative TTE values and no significant differences between the pre-CPB and post-CPB period. Systolic RV echocardiographic parameters decreased post-CPB but this was not accompanied by significant hemodynamic changes.


Introduction
Perioperative right ventricular (RV) dysfunction has been shown to impact outcomes after cardiac surgery. It is associated with increased dose and duration of post-operative vasoactive agent support [1], re-hospitalization [2], short term [3,4] and long term mortality [5]. Global reduction of RV function following cardiopulmonary bypass (CPB) is a well described phenomenon that is associated with increased mortality [1,[6][7][8].
Echocardiography is the main modality for assessing RV function, but quantitative measurement of RV performance is limited by its complex geometry, position, and contraction pattern [9,10]. Validated indices to assess RV systolic function include tricuspid annular plane systolic excursion (TAPSE), fractional areal change (FAC), tricuspid annular velocity (S') and myocardial performance index (MPI) [9,11,12]. However, most indices are validated using transthoracic echocardiogram (TTE) which has limited use compared to transesophageal echocardiogram (TEE) in perioperative settings [12].
The interchangeability between TTE and TEE parameters used for assessing the RV has been investigated in several studies. TAPSE is the most studied variable. TEE M-mode measurements taken in different views have shown conflicting agreement with the standard apical 4 chamber view (AP-4CH) using TTE [13][14][15][16]. Anatomical M-mode TAPSE (AMM-TAPSE) with TEE at the mid-esophageal 4 chamber view (ME-4CH) has shown better agreement than conventional 2D TAPSE measured at ME-4CH compared to the standard TTE TAPSE [13,17]. Other parameters have been less validated and showed fair agreement with measurements obtained with TTE [15,17].
Furthermore, TAPSE and S' have been shown to be reduced after CPB despite unchanged cardiac output (CO) measured by thermodilution [6]. Although TAPSE and FAC have been validated against magnetic resonance imaging (MRI) RV ejection fraction, there are still limited studies on the validation of these echocardiographic parameters with bedside measurement of RV flow and pressure [18].
Our primary objective was to evaluate the agreement of RV TAPSE, FAC, S' and MPI assessed by TEE compared to TTE in cardiac surgical patients. We hypothesized that there will be agreement between TTE and TEE for TAPSE and FAC but less agreement for MPI and S'. The secondary objective was to describe the changes in echocardiographic parameters between the pre-CPB and post-CPB period and evaluate the association of the echocardiographic changes with changes in hemodynamics. We hypothesized that the reduction in echocardiographic parameters after CPB does not reflect a change in RV stroke volume.

Population
This was a single-center, prospective study conducted between January and July 2017 at the McGill University Health Center (MUHC) in Montreal, Quebec, Canada. The study protocol was approved by the MUHC research ethics board (MUHC Study Code: 14-369-SDR). Informed consent was obtained the night prior to surgery. Inclusion criteria were patients > 18 years of age undergoing elective cardiac surgery for whom a perioperative TEE was planned as part of their routine care. Patients who were hemodynamically unstable, had a contraindication to TEE, or had prior sternotomy or thoracotomy were excluded.

Procedure and data collection
The anesthetic procedure for cardiac surgery included induction of anesthesia with a combination of midazolam, propofol, ketamine and rocuronium. Anesthesia was maintained with sufentanil and sevoflurane. Hemodynamic monitoring with a pulmonary artery catheter (PAC) (7.5 Fr Thermodilution Paceport Swan-Ganz catheter, Edwards Lifesciences, Germany) is routinely performed for all cardiac surgery cases at our institution. Patients were ventilated using a volume-controlled ventilation mode with a set tidal volume of 6-8 ml/kg and positive end-expiratory pressure (PEEP) of 5 cmH 2 O.
Echocardiographic measurements were obtained after intubation and after ensuring that the patient was hemodynamically stable. Standard echocardiography machines were used (GE VIVID E95, GE, California, USA) with 6VT-D TEE probes (GE, California, USA) and M5SC-D TTE probes (GE, California, USA). TTE and TEE images were obtained and acquired over 3 consecutive beats in the supine position. All images were acquired and measured by one of two attending cardiac anesthesiologists with extensive echocardiographic experience who are certified in advanced peri-operative TEE and adult echocardiography. Prior to initiation of the study, the protocol, timing, views, and measurements to be obtained were standardized between the two anesthesiologists. TTE was obtained prior to TEE within 10 min of each other, ensuring there were no changes in the patient's hemodynamic profile and no surgical intervention was performed.
The variables of interest were TAPSE, FAC, S' and pulsed wave tissue Doppler (PW TDI-MPI). For TEE measurements, AMM-TAPSE and FAC were measured at the ME-4CH view (Fig. 1A) and S' and PW TDI-MPI were measured at the trans-gastric RV inflow view (Fig. 1B). AMM-TAPSE was obtained by aligning the cursor with the lateral tricuspid annulus then adjusting the angle of the cursor towards the apex and measurement of the maximum difference of excursion was performed using M-mode. (Fig. 1C) FAC was obtained by measuring the end-systolic and end-diastolic area, and the calculation was performed by automated software in the echocardiography machine. PW TDI-MPI was obtained with the pulsed-wave tissue Doppler method by adding the isovolumetric contraction and relaxation times and dividing by the ejection time. (Fig. 1D) All TTE measurements were obtained at the RV focused AP-4CH view (Fig. 1E). Another set of TEE measurements was performed after discontinuation of CPB when hemodynamic parameters were stabilized after sternotomy closure. Normal parameters of RV function were based on the American Society of Echocardiography 2015 recommendations [19].
Inter-observer reliability was assessed by an offline measurement of the obtained images by the other anesthesiologist who did not perform the initial exam and was not aware of the result. This was performed on both pre-CPB TTE and TEE exams on ten patients who were selected randomly.
Stroke volume (SV), measured by intermittent thermodilution via the PAC, was simultaneously assessed during echocardiography prior to CPB and after sternotomy closure. Three cardiac output measurements were recorded, and the average values was used. Central venous pressure (CVP), pulmonary artery pressure (PAP) and heart rate (HR) were also recorded at both time points. Pulmonary artery pulsatility index (PAPi) was calculated with the standard formula. (See supplements).

Statistical analysis
Based on the correlation coefficient for TAPSE from Flo Forner et al., a sample size of 50 would result in a power of > 0.8 [13]. The primary outcome variables are the agreement between TEE versus TTE for each parameter which was assessed by Lin's concordance correlation coefficient (CCC). The strength of agreement was graded with Landis and Koch grading (see supplements) [20]. A Bland-Altman analysis was performed to assess the bias, limit of agreement and precision of the TEE measurements. Inter-rater agreement was assessed with the intra-class coefficient (ICC) [21]. (See supplement for ICC grading).
Mean with standard deviation and median with range was used to describe the hemodynamic and echocardiographic parameters. Differences between pre-CPB and post-CPB was assessed with a paired T-test. The secondary outcome variables were the correlations between hemodynamic variables and echocardiographic measurements which were assessed with the Spearman's rank-order correlation. The level of significance was set at 0.05. All data were analyzed with R statistical software version 4.0.2 [22]. Figures were produced using the package rstatix [23] and ggplot2 [24].

Patient characteristics and baseline echocardiographic and hemodynamic values
Fifty patients were consented and included in the analysis without exclusion. The majority of patients underwent coronary artery bypass graft surgery (CABG) and valvular surgery. Post-CPB, 95.9% required inotropic and/or vasopressor support. (Table 1) Descriptive statistics for pre-CPB and post-CPB echocardiographic and hemodynamic variables are shown in Table 2. Using TTE pre-CPB, 4 patients (8%) had FAC < 35%, 2 patients (4%) had TAPSE < 17 mm, 25 patients (51%) had S' < 9.5 cm/s and none had PW TDI-MPI higher than 0.54. There was good inter-rater reliability in all the measurements obtained. (See supplementary table 1).

Correlation and agreement between pre-CPB TEE and TTE parameters
There was substantial agreement between FAC and AMM-TAPSE pre-CPB while PW TDI-MPI had moderate agreement and S' showed no agreement. The Bland-Altman analysis shows that TEE underestimated S' when compared to TTE due to a significant mean difference of 3.22 cm/s, (95% CI 2.52 to 3.93, p < 0.01). For this measurement, there was poor agreement between TTE and TEE while there was no significant discordance among the other parameters  Table 3). The S' and PW TDI-MPI also shows low precision due to the wide limit of agreement and high (> 30%) percentage error (Fig. 2, Supplemental Figure.

Changes in pre-CPB and post-CPB echocardiographic and hemodynamic variables
Comparison of pre-CPB and post-CPB TEE measurements showed a significant reduction of FAC, S' and AMM-TAPSE, while there was a significant increase in PW TDI-MPI. SV and PAPi remained unchanged post-CPB despite a reduction in echocardiographic indices. Cardiac output increased significantly due to increased heart rate, and there was a significant increase in CVP and mean PAP. ( Table 2).

Discussion
In this cohort of patients undergoing elective cardiac surgery, FAC and AMM-TAPSE acquired at the ME-4CH view had the best (substantial) agreement with the TTE acquired parameters. S' was significantly underestimated by TEE, and PW TDI-MPI showed only moderate agreement. There was significant reduction of AMM-TAPSE and FAC at the post-CPB period suggesting reduced RV function; however, there were no corresponding changes in SV or PAPi.
Amongst the indices for RV systolic function, AMM-TAPSE showed the best agreement with TAPSE acquired with TTE. Anatomic M-mode allows for measurement of TAPSE which represents longitudinal contraction of the RV independent of transducer alignment with the tricuspid annulus. Flo Forner et al. showed that there is good correlation between AMM-TAPSE at the ME-4CH view with TAPSE measurements taken by TTE in a sample size of 30 patients. The agreement at the ME-4CH view was comparable to measurements obtained from the trans-gastric RV inflow view [13]. A subsequent study by Roberts SM et al. found a non-significant mean difference between the two modalities but the agreement was only fair due to the high standard variation in the measurements [17]. This could be due to the inclusion of patients with more variable RV function than in our current study. Our study adds to this pool of evidence that AMM-TAPSE measured at the easily acquired ME-4CH view can be used as a surrogate for TTE measurements with no significant differences between the two modalities.
S' is also another surrogate of RV longitudinal contraction. We found that TEE measurements were significantly lower than TTE measurements with a wide limit of agreement indicating poor accuracy. Previous studies have also shown that TEE significantly underestimates S' in both the trans-gastric RV inflow view and the deep trans-gastric view [16,[25][26][27]. This may be due to the fact that different segments of the RV are interrogated and the alignment of the interrogation angle in the TEE RV inflow view may be more limited than the TTE RV focused view [15].
FAC represents global RV function and we found substantial agreement with TTE measurements without significant bias. However, there was also a wide limit of agreement which may alter some measurements from a normal RV to a dysfunctional state. A study performed in spontaneously breathing patients showed no significant difference between measurements obtained with TEE versus TTE [28]. A subsequent study in anesthetized patients showed only fair agreement between the two modalities [17]. FAC is readily measurable in the ME-4CH view but the accuracy depends on the subjective delineation of the endocardial border. PW TDI-MPI is an index of global RV systolic and diastolic function which requires accurate tissue Doppler or pulsed wave Doppler signal to determine precise timing [9]. High inter and intra observer variability of PW TDI-MPI measurements was found in a previous study which may explain the poor correlation between TEE and TTE measurements in our study [29]. Additionally, the reliability of PW TDI-MPI is decreased when preload is altered which may explain the poor correlation in our study due to the elevated post-CPB CVP [12].
AMM-TAPSE, FAC, and S' were significantly reduced and PW TDI-MPI increased between pre-CPB and post-CPB TEE measurements. Although such changes indicate a decline in RV systolic function, this was not associated with a reduction in SV or PAPi. The finding of reduced longitudinal contraction was also found in previous studies where TAPSE declined by 43%-50% from baseline after CPB [6,30,31]. We found that post-CPB, AMM-TAPSE and S' was moderately associated with CVP and mean PAP. There was a reduction of CVP and mean PAP as AMM-TAPSE and S' increased. We hypothesize that the changes in these parameters may reflect the altered load on the RV during post-CPB more than a change in RV systolic function. Preservation of SV can also be attributed to the use of inotropic agents. Long term implications of echocardiographic reduction of RV function needs to be evaluated. Additionally, TAPSE and S' only represents longitudinal contraction of the RV, SV may be compensated by the increased transverse contraction after CPB [32]. The lack of significant correlations between echocardiographic parameters and SV demonstrated in this study support the role of PAC to provide quantitative measurements of RV cardiac output in addition to echocardiographic findings.
The main limitation of the study is that most of the population had a normal range of echocardiographic and hemodynamic parameters of RV function at baseline. Thus, the robustness of the agreement at extreme values could not be assessed. Secondly, we did not use other novel modalities of echocardiographic assessment such as strain or speckletracking imaging due to the lower reproducibility and applicability at the bedside. Lastly, the measurements were performed in supine, anesthetized and passively ventilated patients in the operating room which may limit the generalizability of the result to patients in the outpatient or intensive care unit setting.

Conclusion
We determined that TEE acquired AMM-TAPSE and FAC had substantial agreement with pre-operative TTE values without significant differences in a population with normal baseline RV function. These two indices could be used to evaluate RV function in the perioperative period. There was significant reduction of RV systolic echocardiographic parameters between the pre-CPB and post-CPB period. However, changes in SV cannot be inferred from changes in echocardiographic indices. Further studies are needed to assess the post-operative outcomes associated with perioperative changes in echocardiographic parameters of RV function.