Pre-implant Right Ventricular dP/dt Can Predict Severe Right Ventricular Failure After Left Ventricular Assist Device Implantation

Right ventricular (RV) failure (RVF) remains a major cause of morbidity and mortality after continuous flow left ventricular assist device (CF-LVAD) implantation, often prolonging a patient’s cardiac intensive care unit stay.¹ Landmark trials with centrifugal flow devices suggest that the incidence of RVF surpasses 30%.^2,3 Post-LVAD RVF is defined by the Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS) as the persistent clinical presentations of RVF after LVAD implantation. Early RVF is further stratified by duration of inotrope or mechanical support requirement into mild (≤7 days), moderate (8–14 days), and severe (>14 days), or the prolonged need for pulmonary vasodilator therapy for >48 hours or direct mechanical support for the RV.¹

Factors contributing to the risk of RVF following CF-LVAD include poor baseline RV contractility or limited contractile reserve, increased perioperative preload, left shift of the interventricular septum leading to RV dilatation and RV dyssynchrony and worsening tricuspid regurgitation (TR).¹ There is no single parameter – either haemodynamic or echocardiographic – that accurately and reliably predicts RVF in patients following CF-LVAD placement.^1,4 While many scoring models have been proposed, their performance is modest when applied to broad populations. ^5–8 In 2018, the European Registry for Patients with Mechanical Circulatory Support (EUROMACS) score was introduced and was suggested to outperform the previous models, with a c-index of 0.7.⁴ Further studies using this score have not yet been published. Moreover, all scoring models rely on multiple clinical and haemodynamic parameters, adding complexity to their clinical use. As such, there is a growing need to develop a tool that can predict early RVF post-CF-LVAD implantation.

The derivative of pressure with respect to time (dP/dt) during early systole is a measurement shown to correlate with myocardial contractility.⁹ Non-invasive measurement of RV dP/dt using the Doppler signal of TR correlates well with invasive dP/dt recordings (Millar catheter – pressure volume loops) and cardiac MRI and has been shown to be associated with outcomes in patients with pulmonary hypertension and for patients undergoing general cardiac surgery.^10–13 However, the association of RV dP/dt with RVF after CF-LVAD has not been assessed.

We aimed to investigate the potential role of non-invasive measurement of RV dP/dt by echocardiogram in predicting early RVF after CF-LVAD implantation. As many candidates for CF-LVAD are inotrope-dependent, we examined both the RV dP/dt with and without inotropy as a potential predictor of post-LVAD RVF.

Methods

Patient Identification

All patients who underwent CF-LVAD placement at Vanderbilt University Medical Center between June 2013 and June 2019 were consecutively evaluated. All methods were reviewed by the Institutional Review Board at Vanderbilt University Medical Center. Patients were included in the study if they had an echocardiogram in the absence of inotropic support performed within 3 months of CF-LVAD placement. Pre-operative demographics, clinical features and invasive right-heart catheterisation measures were obtained from the Vanderbilt ventricular assist device database.¹⁴ For patients who were subsequently treated with an inotrope prior to CF-LVAD placement, on inotrope echocardiogram within 30 days after the baseline echocardiogram was analysed. For individuals with multiple echocardiograms meeting the above criteria, the most recent test prior to LVAD implantation was chosen for analysis.

Outcomes

Based on INTERMACS criteria, post-LVAD RVF was defined by the duration of post-surgical inotrope requirement as mild (≤7 days), moderate (8–14 days), and severe (>14 days), or right ventricular assist device implantation).^1,7 In cases where adjudication was unclear due to peri-operative complexities (i.e. unstable rhythms, deviation from standard post-operative management), two advanced heart failure specialists blinded to the echocardiographic data independently reviewed the electronic medical record and provided a consensus adjudication of the degree of RVF based on INTERMACS criteria. Information about heart transplantation and death after LVAD implantation was collected from the Vanderbilt ventricular assist device database.¹⁴ Overall survival was compared among the subgroups of mild, moderate and severe RVF post-LVAD.

Echocardiogram

Two-dimensional, M-mode and Doppler examinations were performed on iE33 and Cx50 ultrasound systems (Philips). Recordings were re-analysed using clinical post-processing workstations using IMPAX Review Station software (Agfa). The continuous wave Doppler beam was oriented through the tricuspid valve parallel to the TR jet identified by colour Doppler in the apical four-chamber view. Measurements were obtained from TR Doppler profiles that were complete and demonstrated the highest velocity jet. Measurements were made as previously described (Supplementary Figure 1).¹² Briefly, RV dP/dt was evaluated using spectral Doppler recordings from the TR envelope at sweep speeds ranging from 75 to 150 mm/s. The dP/dt measurement was calculated by obtaining the time required from the TR velocity to increase from 0.5 to 2 m/s (i.e. 15 mmHg increase in pressure by Bernoulli’s equation; Supplementary Figure 1), based on prior studies showing a higher degree of correlation with invasively measured dP/dt when measured using this method.¹⁵ All TR Doppler measurements were independently reviewed by two cardiologists blinded to the patient information and outcomes. High concordance of measures between the two observers was noted by linear regression and a Pearson correlation coefficient of 0.90. Other measures of RV function including the tricuspid annular plane systolic excursion (TAPSE), right ventricular fractional area change (RVFAC), and RV free wall tissue Doppler velocity (lateral s’ velocity) were also collected.

Statistical Analysis

Data are presented as mean (SD) or median (interquartile range [IQR]) for continuous/ordinal variables and percentages for categorical/nominal variables. To compare the baseline features among no/mild, moderate and severe RVF post-CF-LVAD, the non-parametric Kruskal–Wallis test by log ranks was used for continuous variables and χ² test (or Fisher’s exact test) was used for categorical variables. Receiver operator characteristic (ROC) curves were constructed to calculate the area under the curve (AUC) and to determine the optimal thresholds for prediction of post-CF-LVAD RVF. Overall survival was estimated by Kaplan–Meier curves and compared by log-rank test among different subgroups. A p-value <0.05 was considered significant. Nominal logistic regression analysis was performed to construct the prediction model with RV dP/dt and other parameters. Statistical analysis was completed using JMP Pro 15 (SAS Institute; http://www.jmp.com) and GraphPad Prism 8 (GraphPad Software; https://www.graphpad.com).

Results

Patient Characteristics

A total of 98 patients met the inclusion criteria. Thirty-three patients did not have an adequate TR Doppler signal on pre-operative echocardiography. While these patients were excluded from our final analysis, an analysis of their baseline characteristics suggests no difference between them and those included in the final analysis. A total of 65 patients were included in the final analysis (Supplementary Figure 2). Among the 65 patients, 40 were started on inotropic therapy prior to LVAD implantation, 32 of whom had measurable on inotrope RV dP/dt on echocardiogram (Supplementary Figure 2). The median age of the patients was 59 years. Most patients (75.4%) were male; 72.3% were white (Supplementary Table 1).

Post-LVAD Severe RVF Predicts Overall Survival

Of the 65 patients, 30 (46.2%), 17 (26.1%) and 18 (27.7%) patients had no/mild, moderate and severe RVF after CF-LVAD, respectively. The baseline clinical characteristics were largely similar among the three subgroups (Supplementary Table 1). During a median follow-up of 304 days (IQR 135, 2,078), 33 (50.8%) patients underwent heart transplantation and 23 (35.4%) patients died. Twelve patients died within the first 6 months after LVAD implantation. Patients with severe post-LVAD RVF were less likely to undergo heart transplantation (63.3% versus 58.8% versus 22.2% in the no/mild, moderate, and severe RVF subgroups; p=0.017). There was a significant difference in the Kaplan–Meier overall survival curves within 6 months after LVAD implantation between the no/mild, moderate and severe RVF subgroups (p=0.0005; Figure 1). A significantly higher mortality within 6 months after LVAD implantation was observed in patients with severe RVF (severe RVF versus no/mild RVF HR 9.6; 95% CI [2.0–44.6]; p=0.004, severe RVF versus moderate RVF HR 5.0; 95% CI [1.1–23.4]; p=0.038), while there was no significant difference between no/mild and moderate RVF (moderate RVF versus no/mild RVF HR 1.9; 95% CI [0.3–13.5]; p=0.52) (Figure 1).

There was a significant difference in the Kaplan–Meier overall survival curves during the entire follow-up period after LVAD implantation between the three RVF subgroups (p=0.0044) as well as a similar trend when comparing RVF subgroups (severe RVF versus no/mild RVF mortality HR 3.8; 95% CI [1.5–9.7]; p=0.006, severe RVF versus moderate RVF HR 3.6; 95% [1.2–11.6]; p=0.038, moderate RVF versus no/mild RVF HR 1.0; 95% CI [0.3–3.5]; p=0.95; Supplementary Figure 3).

The cohort was subsequently divided into those who developed post-CF-LVAD severe RVF and those who did not (none/mild plus moderate). The demographic characteristics and baseline clinical features were largely similar between the patients with severe RVF and the rest of the cohort (Supplementary Table 2). There were no differences between the two subgroups in the proportion of patients requiring inotropic therapy prior to LVAD (Supplementary Table 2).

Baseline Echocardiographic RV dP/dt is Associated with Severe RVF after CF-LVAD

We first evaluated whether the baseline echocardiographic measures of RV function without inotropes are associated with outcomes of RV failure after CF-LVAD placement. In this cohort of 65 patients, the median duration between the off inotrope echocardiogram and LVAD implantation was 16 days (IQR 7, 33). RV dP/dt in the severe RVF subgroup was significantly lower than that of the rest of the cohort (median 248 versus 281 mmHg/s; severe RVF versus no severe RVF group; p=0.022; Table 1 and Supplementary Figure 4A). In contrast, there were no significant differences in baseline TAPSE, RVFAC or lateral wall S’ (Table 1 and Supplementary Figures 4B–4D). The ROC curve identified the optimal cut-off for RV dP/dt to be 300 mmHg/s (sensitivity 88.9%; specificity 44.7%; negative predictive value [NPV] 91.3%; positive predictive value 38.1%; AUC 0.685; p=0.014; Figure 2).

In our cohort of 65 patients, 60 had an off inotrope right heart catheterisation procedure prior to LVAD. The median duration between the right heart catheterisation and LVAD implantation was 11 days (IQR 6, 24). Haemodynamic parameters for assessing RV function, such as central venous pressure (CVP), pulmonary arterial pulsatility index (PAPi) and CVP:pulmonary capillary wedge pressure (PCWP) ratio, did not significantly differ between the two groups (Supplementary Table 3). Due to the retrospective nature of the study, on inotrope haemodynamic data – often recorded by the bedside – were largely not available for analysis.

RV dP/dt On Inotrope Identifies Patients with Contractile Reserve Prior to CF-LVAD Placement

To further assess whether RV dP/dt serves as a valuable adjunct tool in identifying the subset of patients with RV contractile reserve, we evaluated RV dP/dt in the subset of patients started on inotropes prior to CF-LVAD placement. Forty patients were identified to have inotrope initiation prior CF-LVAD placement, 32 of whom had interpretable TR Doppler waveforms (Supplementary Figure 2). The median duration between the off inotrope and on inotrope echocardiograms was 4 days (IQR 2, 6).

Of the 32 patients, 18 (56%) had a baseline RV dP/dt <300 mmHg/s and 14 had a baseline RV dP/dt ≥300 mmHg/s. Only two of the 14 patients with off inotrope RV dP/dt ≥300 mmHg/s developed severe RVF post-LVAD, echoing our finding that a baseline RV dP/dt ≥300 mmHg had a high NPV for severe RVF post-LVAD. In contrast, seven of the 18 patients with a baseline RV dP/dt <300 mmHg/s developed post-LVAD severe RVF, while 11 patients did not. Among these 18 patients, inotrope initiation resulted in an increase of dP/dt to ≥300 mmHg/s in seven (‘inotrope responders’), only one of whom had post-LVAD severe RVF after CF-LVAD placement. Of the 11 patients who had persistently low RV dP/dt <300 mmHg/s despite inotropes (inotrope non-responders), seven (63.6%) had severe RVF post-LVAD. The inotrope non-responders had a significantly higher likelihood of developing post-LVAD severe RVF compared with the other on inotrope patients (67% versus 14.3%; OR 10.5; 95% CI [1.8–59.4]; p=0.013; Figure 3).

Other echocardiographic parameters including off inotrope TAPSE, lateral S’ and RVFAC, were not predictive of post-LVAD severe RVF either in patients off or on inotrope (data not shown).

Discussion

Our study demonstrates that RV dP/dt can serve as a surrogate marker for RV contractility and a predictor of severe RV failure acutely post LVAD implantation. RV dP/dT is an easily accessible bedside tool that could provide a valuable pre-LVAD assessment to perform pre-implantation to help optimise post-implantation outcomes.

A significantly worse outcome was observed in patients with severe RVF post-LVAD compared with no, mild and moderate RVF, while there was no significant difference in mortality between no/mild versus moderate RVF. Our findings are corroborated by a previous study by LaRue et al. and suggest the importance of identifying patients with post-LVAD severe RVF.¹⁶ A mean RV dP/dt value of 1,016 mmHg/s was previously reported in healthy subjects, while a reduced baseline RV dP/dt <410 mmHg/s was associated with poor 1-year outcome in patients with pulmonary arterial hypertension and chronic thrombo-embolic pulmonary hypertension.¹² In comparison, the median level of dP/dt in patients in our study was 261 mmHg/s, reflecting the severity of RV dysfunction (often under-appreciated) in advanced heart failure patients.

In our study, a baseline off inotrope RV dP/dt ≥300 mmHg/s predicted a low risk of post-LVAD severe RVF (NPV 91%). Furthermore, among the patients with off inotrope dP/dt <300 mmHg/s who were started on inotropes prior to LVAD, only one of the seven patients (14.3%) who had improved on inotrope RV dP/dt ≥300 mmHg/s developed severe RVF post-LVAD, while seven of the 11 patients (67%) with both off inotrope and on inotrope RV dP/dt <300 mmHg/s developed severe RVF, suggesting a role for RV contractile reserve evaluation with an inotrope challenge in the further risk stratification of post-LVAD severe RVF. Moreover, as baseline off inotrope RV dP/dt might not always be available for inotrope-dependent patients at the time of LVAD evaluation, we propose that a unified threshold of RV dP/dt of 300 mmHg/s for severe RVF prediction can be of value for clinical application, regardless of the inotrope status. In light of the high sensitivity of this cutoff, it is important to note that not all patients meeting this threshold will experience severe RVF post-LVAD. Hence, it may not be considered as a contraindication to LVAD implantation but rather help guide peri-operative care among those at higher risk of developing post-LVAD RVF.

Previous echocardiographic attempts to predict RVF have poor accuracy, consistent with our findings.¹ RV geometry is often not amenable to volumetric assessment by 2D echocardiography; TAPSE measurement assumes base-to-apex shortening rather than radial forces, which correlates with RV stroke volume.¹¹ Haemodynamic assessments, including CVP, CVP:PCWP >0.6 and PAPi, have been widely used for the prediction of RVF, but to date, no single measure is consistently predictive across studies.^1,17 This is likely, in part, explained by the dependence of the haemodynamic parameters on preload and afterload. However, unlike these other parameters, RV dP/dt is a simple and reproducible tool that is widely available and inexpensive. For most patients, the ability to predict the absence of severe RVF is paramount, making the high NPV of dP/dt particularly important. Several predictive models have been proposed but each has only demonstrated modest discrimination at best.¹⁸

Given these findings, we propose an algorithm (Figure 4) for using RV dP/dt in predicting post- LVAD severe RVF. A baseline off inotrope RV dP/dt ≥300 mmHg/s predicts a low risk of severe RVF after LVAD implantation and a favourable long-term prognosis. For patients with a baseline RV dP/dt <300 mmHg/s (or for patients who cannot have an off inotropic study), an on inotrope RV dP/dt of ≥300 mmHg/s also predicts a low risk of severe RVF. On the other hand, having both off inotrope and on inotrope RV dP/dt <300 mmHg/ml/m² carries a high likelihood of post-CF-LVAD severe RVF.

Limitations

There are several limitations to our study. First, it is a retrospective analysis of a single centre with a small cohort. Our cohort number was largely limited by the criterion requiring a patient to have off inotrope echocardiogram within 3 months prior to CF-LVAD implantation and availability of the echocardiogram studies for manual RV dP/dt measurement. Second, the clinical use of RV dP/dt cannot be applied to patients without a good TR envelope detected by echocardiogram. Third, while RV dP/dt with a cut-off of 300 mmHg/s offers a good sensitivity for detecting post-LVAD severe RVF, the specificity is modest. Our data suggest that inotrope challenge for patients with an off inotrope RV dP/dt <300 mmHg/s could be helpful for further risk stratification. Fourth, only 20% of our cohort included patients with HeartMate III; it is plausible that device-specific risk factors, such as artificial pulsatility and pump speed, may affect the incidence of RVF post-implantation. Further prospective studies are needed to confirm this finding. Fifth, preoperative echocardiographic assessment of RV diastolic function could not be performed in our analysis. Lastly, RV dP/dt, while reflective of contractility, may be influenced by factors such as preload, RV geometry and arrhythmias. While echocardiograms were performed in relatively stable preoperative settings, we could not directly quantify or control for hepatic congestion, RV size or filling pressures at the time of echocardiogram, which introduces some variability into the dP/dt measurements.

Conclusion

Echocardiography-derived RV dP/dt – regardless of inotrope use status – strongly associates with post-CF-LVAD RVF. A baseline or on inotrope RV dP/dt of ≥300 mmHg/s identifies patients at low risk for early post-LVAD severe RVF, while the presence of both off inotrope and on inotrope RV dP/dt <300 mmHg/s carries a higher risk of severe RVF post-LVAD implantation. Our findings should be further evaluated in future prospective studies, in particular regarding how they pertain to various sub-groups.

Click here to view Supplementary Material.