Post-operative Right Ventricular Failure After Cardiac Surgery: A Cohort Study

Introduction: Right ventricular failure (RVF) after cardiac surgery is an important risk factor for morbidity and mortality. Its diagnosis is challenging, and thus, its incidence and predictors are not well-established. We investigated the incidence, complications, and variables associated with clinically relevant post-operative RVF. Methods: We included all patients who underwent cardiac surgery with cardiopulmonary bypass between 2016 and 2019 in a cardiac surgery center with standardized diagnostic and therapeutic management of RVF. RVF was considered only if clinically relevant: associated with hemodynamic instability requiring catecholamine support and inhaled nitric oxide relayed by sildenafil. Results: Overall, 3,826 patients were included, of whom, 110 (2.9%) developed post-operative RVF. Mortality was not different among patients who developed post-operative RVF, compared with the rest of the cohort (1.8 vs. 0.7%, p = 0.17). Using a composite outcome that combined death, reintubation, stroke, and prolonged intensive care unit stay (more than 14 days) yielded an incidence of 6.6%, and RVF was associated with this composite outcome with an odds ratio of 3.6 (2.2–5.8), p < 0.001. In a multivariable model, pre-operative variables independently associated with post-operative RVF were pre-operative atrial fibrillation (AF) {adjusted odds ratio (adjOR) 3.22 [95% confidence interval (95%CI) = 1.94–5.36], p < 0.001}, left ventricle ejection fraction below 50% [adjOR = 2.55 (95%CI = 1.52–4.33), p < 0.001], systolic pulmonary artery pressure above 55 mmHg [adjOR = 8.64 (95%CI = 5.27–14.1); p < 0.001], mitral valve surgery [adjOR = 2.17 CI (95%CI = 1.28–3.66), p = 0.004], and tricuspid valve surgery [adjOR = 10.33 (95%CI = 6.14–17.4), p < 0.001]. In patients who developed post-operative RVF requiring treatment, 32 (29.1%) showed RV dysfunction before surgery. Conclusion: In this cohort study, 2.9% of patients developed clinically significant post-operative RVF. Moreover, RVF was associated with severe adverse outcomes, including death, strokes, reintubation, and prolonged intensive care unit stay.


INTRODUCTION
Right ventricular failure (RVF) after cardiac surgery has been found as a significant cause of morbidity and mortality in several settings (1). Mechanisms of RVF include suboptimal myocardial protection during surgery, long cardiopulmonary bypass (CPB) time, right ventricular (RV) myocardial ischemia or infarction, atrial arrhythmias or loss of atrioventricular synchrony, reperfusion lung injury with secondary pulmonary hypertension (PH), post-operative pulmonary micro-or macroembolism, pre-existing pulmonary vascular disease, protamineinduced PH, and others (2). An acute increase in PH leads to higher RV afterload, altered RV-pulmonary artery coupling, and then RV failure with compromised hemodynamic (3). Pulmonary vascular resistance (PVR) may increase after CPB because of atelectasis, ischemia-reperfusion injury with inflammatory cytokines, endothelial damage, and reduction of nitric oxide release (4,5).
Post-operative RVF management includes pulmonary vasodilators, optimal fluid management, inotrope support, and, when necessary, restoring a compromised RV myocardial perfusion pressure, systemic vasopressors. As of yet, these treatments are not supported by robust studies. Inhaled nitric oxide (iNO) has been approved for persistent PH of the newborn but is used off-label in cardiac post-operative RV failure despite poor evidence in hard clinical outcomes (6)(7)(8).
Similarly, sildenafil showed promise in clinical studies, in post-operative care, by reducing pulmonary pressures and resistance with no alteration in systemic hemodynamic and oxygenation (9,10).
Another issue lies in the lack of a consensual definition of RVF after surgery (11). Consequently, the incidence of RVF and its risk factors are not well-established. Although some reports described risk scores of RVF after left ventricular assist device implantation, the literature is poor in standard cardiac surgery (12,13).
This study aimed to characterize the incidence, outcomes, and predictors of RVF after cardiac surgery.

Study Population and Design
From January 1, 2016, to December 31, 2019, the study included all adult patients undergoing cardiac operations with CPB. The exclusion criteria were age below 18 years. We performed a retrospective analysis of data collected prospectively, including perioperative characteristics, postoperative drugs, and outcomes. Data were anonymized as per national regulation and used with the approval of an institutional review board committee (Commission Nationale de l'Informatique et des Libertés, CNIL, declaration number 2109982, 17/10/2017). All the data are part of the Registry for the Improvement of Post-operative Outcomes in Cardiac and Thoracic Surgery database (NCT03209674). Patients' opposition to the use of anonymized data by investigators was systematically sought (i.e., informed consent was obtained from all patients).

Definitions
Post-operative RVF was defined by the combination of (1) hemodynamic instability requiring administration of vasopressors and/or inotropes and (2) immediate postoperative needs of pulmonary vasodilators (iNO followed by sildenafil R per os). In the post-operative period, patients were systematically explored with echocardiography when presenting with hemodynamic instability. RVF required hypokinesia of the RV free wall or RV/left ventricle (LV) interdependence or enlarged RV with RV/LV ratio > 1. Additional optional parameters included tricuspid annular plane systolic excursion (TAPSE) < 16 mm and tissue doppler imaging with peak myocardial velocity at the lateral tricuspid annulus (S wave) < 9.5 cm/s, also considered if patients did not undergo tricuspid intervention.

Right Ventricular Failure Management
Inhaled NO was initiated at RVF diagnosis. It was administered at 10 ppm and once hemodynamic stability was achieved for at least 24 h; weaning was achieved by decreasing 1 ppm every hour. Sildenafil was initiated during the first 48 h after surgery, whenever oral intake was possible, with a dosage of 20 mg every 8 h.

Statistical Analysis
Data distribution was assessed using the Kolmogorov-Smirnovtest with Lilliefors correction. Continuous variables were summarized as mean ± standard deviation or median (interquartile range), as appropriate, and were compared, using the Student t-test or Mann-Whitney U-test accordingly. Categorical variables were presented as count (percentage) and compared using the chi-square or Fisher exact-test, as appropriate.
Binary logistic regression analysis was performed to assess variables independently associated with RVF. The multivariable model was built by the stepwise selection, with an entry criterion of p < 0.10. Variables included in the model were age, sex, previous AF, estimated glomerular filtration rate, systolic pulmonary arterial pressure (sPAP) > 55 mmHg, left ventricular ejection fraction (LVEF) < 50%, mitral surgery, tricuspid intervention, and CPB length.
Comparing patients who presented post-operative RVF and those who did not, regarding mortality, was performed using proportion comparison, and using survival analysis, censored at 14 days, with Kaplan-Meier estimates. This censorship was used because 97% of patients were discharged from the intensive care unit (ICU) by day 14, and the remaining 3% presented heterogeneous length of stay, precluding from linear assumptions. For sensitivity, uncensored analyses were performed.
All statistical analyses were performed using IBM SPSS version 26.0 (IBM Corporation, Armonk, New York). A twosided p-value < 0.05 was considered to be statistically significant. Figures were drawn using Prism 8.4.3 (GraphPad Software, California, USA) and R software 3.6 (R project, worldwide community software).

Patients and Baseline Characteristics
We included 3,826 consecutive patients who underwent cardiac surgery with CPB. Baseline characteristics are shown in Table 1.
Focusing on the 14-day in-hospital follow-up, patients who developed post-operative RVF were less likely to be discharged at this point than patients who did not present RVF (log-rank p < 0.0001, see Figure 2). The uncensored analysis yielded a similar outcome between patients with or without post-operative RVF (log-rank p = 0.70). For sensitivity, we computed a composite outcome that combined death, reintubation, stroke, and prolonged ICU stay (more than 14 days). We found an incidence of 252/3,826 (6.6%), and RVF was associated with this composite outcome (19.1 vs. 6.2%, odds ratio = 3.6 (2.2-5.8), p < 0.001).

Variables Associated With the Incidence of Right Ventricular Failure
In a logistic regression multivariable model, variables independently associated post-operative RVF were preoperative AF {adjusted odds-ratio (adjOR) 3. 22 Table 3).

Pre-operative Right Ventricular Function in Patients With Post-operative Right Ventricular Failure
Focusing in patients who presented RVF (n = 110), preexisting RV dysfunction [with either TAPSE < 16 mm and tissue doppler imaging with peak myocardial velocity at the lateral tricuspid annulus (S wave) < 9.5 cm/s] was found in 32 (29.1%) patients ( Table 4). Compared with those with preserved pre-operative RV function, those with pre-operative

DISCUSSION
In this study, we described the incidence, risk factors, and prognosis of post-operative RVF. We observed that (1) clinically significant RVF after cardiac surgery is rare, occurring below 3%; (2) despite higher predicted perioperative risk (i.e., EuroScore II), treatment with iNO and sildenafil allowed to reach similar post-operative results regarding hard clinical outcomes when comparing patients with post-operative with and without RVF; (3) independent variables associated with post-operative RVF included the history of AF, pre-operative LVEF below 50%, mitral surgery, tricuspid surgery, and pre-operative sPAP above 55 mmHg; (4) in patients who presented post-operative RVF, its severity was associated with pre-operative RV dysfunction. Post-operative RVF in cardiac surgery ICU is a major concern because of its challenging diagnosis: sternotomy is an obstacle to procure accurate and reproducible information regarding echocardiographic right heart evaluation. Due to the standard approach in cardiac surgery with pericardial opening, TAPSE may be inherently reduced in the post-operative period without patent clinical RVF (14). Moreover, TAPSE after tricuspid annuloplasty may lead to an underestimation of RV systolic function (15). Definition of RVF is heterogeneous among physicians in the field and the literature, going from only echocardiographic definitions to mild or very severe circulatory failure (16,17). Although echocardiography is an overall powerful tool, in the post-operative setting, it presents limitations, the first of which is a possible lack of sonographic window allowing for accurate measurements, particularly in mechanically ventilated patients (18). Meanwhile, a single echocardiographic parameter may not be sufficient to define RVF, and as echocardiographic assessment of longitudinal RV systolic function such as TAPSE and S wave may be eyeballed by expert visual assessment (19), we chose a pragmatic endpoint defined as echocardiographic assessment RVF combined with clinical hemodynamic instability requiring immediate treatment.
Our data could not allow for the analysis of the effect of iNO and sildenafil, as RVF treatment, because of a lack of a control group. Recent studies showed interest in sildenafil administration on its own or in association with iNO. In our experience, combining both drugs in the post-operative period seemed to be safe and efficient in decreasing both pressure and resistance in the pulmonary circulation without affecting cardiac output, although, as of yet, no study showed differences in clinical outcomes (20)(21)(22).
We confirmed that mitral surgery and previous LVEF < 50% were associated with post-operative RVF, as already described previously, with a high incidence of low cardiac output state after mitral valve procedures (23,24). The principal mechanism invoked is an abrupt increase in afterload due to elimination of the reverse flow into the left atrium, all the more so if LV was decreased before correction (25,26).
As we observed, previous AF could be associated with a complicated post-operative course (27). Gorter et al. showed in patients with heart failure with preserved LVEF that RV and RA functions were more depressed in patients with AF and patients in sinus rhythm but with prior AF, compared with patients without any history of AF (28). Also, they suggested these findings were independent of pulmonary pressures. According to recent studies, functional TR is associated with right atrium (RA) enlargement in the setting of chronic AF (29). Thus, AF, tricuspid regurgitation, and RA enlargement are somehow working together, participating in the development of RVF. Then, TR correction could represent a key therapeutic option, but it has shown to be a major risk factor for post-operative RVF (30,31). Before surgery, the RV adapted to chronic volume overload, but surgical TR correction could lead to an abrupt increase in afterload with reduced pre-load and then uncoupling RV and pulmonary circulation as an analogy with LV hemodynamics in mitral valve closure, but because of RV anatomy and physiology, it would not support such hemodynamic modification as the LV could accept it. Besides, CPB with induced RH and abnormal septal motion has shown deleterious effect in RV hemodynamics (2). Finally, severe pre-operative RH is a well-known risk factor for post-operative RVF, corresponding to our findings with a threshold of sPAP > 55 mmHg in our study. We acknowledge that this study does not solve the issue of whether iNO and sildenafil are appropriate to treat RVF, but it comforts the fact that RVF may be anticipated in some procedures (mainly tricuspid and mitral valve surgery) and that these treatments seem to alleviate the poor prognosis of an RVF outcome.
We acknowledge several limitations. First, although we chose criteria that satisfied clinician need for post-operative RVF (i.e., which required specific treatment, including vasopressors and inhaled NO), there are no consensual and easily obtainable RVF criteria in clinical practice. Secondly, lack of difference in mortality may be imputed to a lack of power; all the more so that the computation of a composite outcome which combined deaths, strokes, prolonged ICU stay, and reintubation showed significantly increased risk in patients who developed post-operative RVF. Third, our center only performed standard cardiac surgery excluding congenital cardiac procedures, heart transplantation, or LVAD implantations, all considered at high risk for post-operative RVF. Fourth, pre-operative RV assessment parameters were only retrievable in patients who developed post-operative RVF; thus, we could not integrate these elements into multivariable analysis regarding pre-operative risk factors of developing post-operative RVF. Nevertheless, we managed to study pre-operative RV function in patients with post-operative RVF, providing important insight into this subgroup in favor of a link between RV dysfunction, AF, TR, and enlarged RA.
Finally, we acknowledge that the lack of a control group (i.e., patients who presented post-operative RVF but were treated with neither iNO nor sildenafil) makes any analysis of treatment efficacy impossible.
Thus, we advocate for the design of a dedicated randomized controlled trial to address this rare but severe issue that is postoperative RVF.

CONCLUSION
Post-operative RVF occurred in 2.9% of patients after cardiac surgery. Variables significantly associated with RVF were history of AF, LVEF < 50%, sPAP > 55 mmHg, and mitral and tricuspid valve procedures.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. The ethics committee waived the requirement of written informed consent for participation.

AUTHOR CONTRIBUTIONS
DLe wrote the initial manuscript, participated to data collection, and performed statistical analyses. DLa, PE, AB, and PS participated to data collection and provided critical review. AB also curated the database. LN supervised the study, participated to data collection, analyses, and wrote the final manuscript. All authors contributed to the article and approved the submitted version.