Author + information
- Received October 30, 2017
- Revision received December 18, 2017
- Accepted December 21, 2017
- Published online April 1, 2019.
- Masahiko Asami, MDa,
- Stefan Stortecky, MDa,
- Fabien Praz, MDa,
- Jonas Lanz, MDa,
- Lorenz Räber, MD, PhDa,
- Anna Franzone, MDa,
- Raffaele Piccolo, MDa,
- George C.M. Siontis, MD, PhDa,
- Dik Heg, PhDb,
- Marco Valgimigli, MD, PhDa,
- Peter Wenaweser, MDa,
- Eva Roost, MDc,
- Stephan Windecker, MDa and
- Thomas Pilgrim, MDa,∗ ()
- aDepartment of Cardiology, Swiss Cardiovascular Center, Bern University Hospital, Bern, Switzerland
- bInstitute of Social and Preventive Medicine and Clinical Trials Unit, University of Bern, Bern, Switzerland
- cDepartment of Cardiac Surgery, Swiss Cardiovascular Center, Bern University Hospital, Bern, Switzerland
- ↵∗Address for correspondence:
Dr. Thomas Pilgrim, Department of Cardiology, Swiss Cardiovascular Center, Bern University Hospital, Freiburg Strasse 10, CH-3010 Bern, Switzerland.
Objectives The purpose of this study was to investigate the association between right ventricular dysfunction (RVD) and cardiovascular death after transcatheter aortic valve replacement (TAVR).
Background There is conflicting evidence on the effect of RVD on clinical outcomes after TAVR.
Methods A total of 1,116 TAVR patients (age 82 ± 6 years; 51% female) who were consecutively enrolled into a prospective registry underwent detailed pre-operative assessment of right ventricular (RV) function and were dichotomized into 2 groups for the purposes of the present retrospective analysis. RVD was assessed using fractional area change (<35%), tricuspid annular plane systolic excursion (<1.7 cm), and systolic movement of the RV lateral wall by tissue Doppler imaging (<9.5 cm/s). RVD was found in 325 (29.1%) patients. The primary outcome was cardiovascular death at 1 year.
Results After adjustment for comorbidities, patients with RVD had a higher risk of cardiovascular death at 1 year compared with patients with normal RV function (20.1% vs. 7.1%; adjusted hazard ratio [HRadj]: 2.94; 95% confidence interval [CI]: 2.02 to 4.27; p < 0.001). The difference emerged within the first 30 days after TAVR (9.0% vs. 2.2%; HRadj: 4.62; 95% CI: 2.51 to 8.50; p < 0.001). Normalization of RV function after TAVR was found in 57.4% of patients with RVD at baseline. There was a gradient of increasing risk of cardiovascular death among patients with normal RV function, RVD recovery (HRadj: 2.16; 95% CI: 1.16 to 4.02), new RVD (HRadj: 3.93; 95% CI: 2.09 to 7.39), and maintained RVD (HRadj: 8.74; 95% CI: 5.33 to 14.3), respectively.
Conclusions RVD at baseline was associated with a more than 2-fold increased risk of cardiovascular death at 1 year after TAVR, with a gradient of risk according to RVD recovery. (Swiss TAVI Registry; NCT01368250)
- aortic stenosis
- recovery of right ventricular function
- right ventricular function
- transcatheter aortic valve replacement
In patients with severe aortic stenosis (AS), chronic pressure overload in the left ventricular chambers can be transmitted through the pulmonary vascular system and result in compensatory right ventricular (RV) remodeling, dilatation, and eventually right ventricular dysfunction (RVD) (1). RVD has been documented in up to 1 in 4 patients with severe AS (2,3), and may occur even more frequently among patients with low-flow low-gradient AS (1).
RVD has been associated with adverse outcome after cardiac surgery (4). In contrast, findings from studies investigating the effect of RVD on clinical outcomes in patients undergoing transcatheter aortic valve replacement (TAVR) have revealed conflicting results. In a subanalysis of the PARTNER (Placement of Aortic Transcatheter Valves) II trial (inoperable cohort), an association of baseline RVD and mortality at 1 year was no longer significant after multivariable adjustment (5). Similarly, there was no significant difference in survival in TAVR patients with or without RVD in a single-center cohort of more than 600 patients (3). Conversely, RVD was identified as an independent predictor of late mortality in a report by Griese et al. (6). However, quantitative measures of RV assessment have not been consistently applied in previous studies and the numbers of patients with RVD were modest. We therefore aimed to investigate the association of quantitatively assessed RVD according to current American and European guidelines with clinical outcome after TAVR.
All patients that undergo TAVR at Bern University Hospital, Bern, Switzerland, are consecutively included in a prospective registry as part of the Swiss TAVI Registry (NCT01368250). For the purpose of the present retrospective analysis of prospectively acquired echocardiographic data, we considered all patients with available transthoracic echocardiography within 3 months before the intervention and at time of hospital discharge after TAVR. Device selection and all decisions on periprocedural management were left to the discretion of the operators. Patients were excluded from the analysis if no transcatheter heart valve was implanted, if a non–CE-marked device was used, if no echocardiography was available within 3 months of the procedure, and in case of a poor acoustic window not allowing for a reliable assessment of fractional area change (FAC) in a focused apical 4-chamber view, tricuspid annular plane systolic excursion (TAPSE) by M-mode between end diastole and peak systole, or DTI in an apical view with parallel alignment of the Doppler beam with the RV free wall longitudinal excursion. All data were prospectively entered in a web-based database held at the Clinical Trial Unit at the University of Bern, Switzerland. All patients were prospectively followed at regular intervals. The Bern TAVI registry has been approved by the local ethics committee. Written informed consent was obtained for participation in this registry from all patients.
Assessment of RV function
All subjects underwent transthoracic echocardiography with a Philips iE33 machine (Philips Healthcare, Andover, Massachusetts) within 3 months before TAVR and within 1 month after TAVR. All echocardiographic studies were performed by a board-certified cardiologist and evaluated by an independent second reader (M.A.) trained in echocardiography. For the ascertainment of echocardiographic parameters, measurements from at least 3 consecutive heartbeats were averaged. Echo loops were analyzed at a workstation allowing for offline analysis (Syngo Dynamics Workplace, version 9.5, Siemens Medical Solutions, Malvern, Pennsylvania). According to the current guidelines of the American Society of Echocardiography (ASE)/European Association of Cardiovascular Imaging (EACVI) (7) and a statement from the Heart Failure Association and the Working Group on Pulmonary Circulation and Right Ventricular Function of the European Society of Cardiology (8), presence or absence of RVD is pre-defined by the following parameters of RV function and dimension: FAC, TAPSE, and systolic movement of the RV lateral wall using tissue Doppler imaging (Sʹ). Cutoffs for abnormal RV function were as follows: FAC ([RV end-diastolic area − RV end-systolic area]/RV end-diastolic area × 100) <35%, TAPSE <1.7 cm, and Sʹ <9.5 cm/s. RVD was present if >50% of the available RV function parameters were below the lower cutoff value (Figure 1). The analysis was repeated using a hierarchical evaluation of RV function parameter that was primarily categorized by TAPSE <1.7 cm, then by Sʹ <9.5 cm/s if TAPSE was unavailable; if both TAPSE and Sʹ were unavailable, FAC <35% was used to determine RV function status. All 3 parameters for characterization of RV function were available in 404 subjects. Inter-rater and intrarater variability of RVD assessment showed very good agreement in a subset of 50 patients (Supplemental Table 1).
TAVR was performed according to standard techniques (9). A transfemoral approach was used by default unless prohibited by peripheral vascular disease, and was performed under local anesthesia and conscious sedation. TAVR by alternative approach was performed under general anesthesia. Post-procedural care comprised heart rhythm monitoring for at least 48 h after intervention, laboratory testing, and 12-lead ECG directly after the procedure and then on a daily basis, with echocardiographic control before discharge.
Clinical follow-up and endpoint assessment
After hospital discharge, clinical follow-up was performed at 30 days and 1 year after TAVR. For the ascertainment of clinical endpoints, standardized interviews, documentation from referring physicians, and hospital discharge summaries were used. All adverse events were systematically collected and independently adjudicated according to the updated definitions of the Valve Academic Research Consortium-2 (10). The primary pre-specified endpoint of our analysis was cardiovascular death within 1 year after TAVR. Secondary endpoints included all-cause death, major adverse cardiac and cerebrovascular events (MACCE), disabling stroke, and periprocedural myocardial infarction at 30 days and 1 year after TAVR. MACCE was a composite of all-cause death, disabling stroke, and myocardial infarction. Life-threatening and major bleeding, kidney injury (stage 3), and major access site complications were evaluated at 30-day follow-up. New York Heart Association (NYHA) functional classification was assessed at baseline, 30 days, and 1 year after TAVR.
Clinical presentation and procedure
Baseline, echocardiographic, and procedural data are presented as mean ± SD (with p values from Student t tests for 2-group comparisons, or F-test for comparisons of more than 2 groups) or as counts with percent of patients (with p values from Fisher exact test for 2-by-2 comparisons or chi-square tests), comparing RVD with normal RV function.
Clinical endpoints (death, cardiovascular death, myocardial infarction, cerebrovascular events, bleeding, kidney injury, and access site complications) are reported as counts of the first occurrence of the event within 30 days or 1 year of follow-up (% of from life table estimates), comparing the 2 groups using Cox regression (i.e., censoring patients at death or who were lost to follow-up). Reported are crude hazard ratios (HRs) with 95% confidence intervals (CIs) (with p values from Wald chi-square tests, or continuity correct risk ratios with p values from Fisher exact tests in case of zero events). Similarly, maintained RVD, new RVD, and recovery of RV function were compared pairwise against maintained normal RV function. Reported is adjusted hazard ratio (HRadj) with 95% CI, with normal RV function as the reference category, adjusted for sex, history of cerebrovascular events, diabetes, and logistic EuroSCORE ≥40%. (These variables were selected with p < 0.10 effect in a multivariable model to predict MACCE at 1 year.) Adjusted analyses were conducted in case 10 events or more were available overall.
Predictors of clinical endpoints
Predictors of 1-year MACCE, all-cause death, and cardiovascular death were separately evaluated (Supplemental Tables 2 and 3) including RVD as main effect, age, diabetes, peripheral vascular disease, arterial hypertension, body mass index ≤20 kg/m2, chronic obstructive pulmonary disease, previous stroke or transient ischemic attack, logistic EuroSCORE ≥40%, moderate or severe post-aortic regurgitation (AR), sex, atrial fibrillation, NYHA functional class III or IV, concomitant percutaneous coronary intervention, coronary artery disease, and creatinine >200 μmol/l (univariable Cox regressions, if p < 0.10 is added to multivariable Cox’s regression model) (11–13).
Analyses were performed with Stata version 14.2 (StataCorp, College Station, Texas). Two-sided p values <0.05 were considered statistically significant.
The study flow diagram is shown in Figure 2. Among 1,339 consecutive patients undergoing TAVR between August 2007 and December 2015 at Bern University Hospital, 1,116 had raw echocardiographic data of adequate quality for the assessment of RVD and represent the study population of the present analysis. RVD was found in 325 (29.1%) patients. Compared with patients with normal RV function, patients with RVD at baseline were younger (81.3 ± 7.1 years vs. 82.5 ± 5.5 years; p = 0.002), were more often male (56.6% vs. 45.8%; p = 0.001), more commonly had a history of previous coronary artery bypass grafting (19.4% vs. 9.3%; p < 0.001), atrial fibrillation (47.7% vs. 29.3%; p < 0.001), or renal failure (76.2% vs. 68.8%; p = 0.01), and were more likely to present with NYHA functional class III or IV (76.2% vs. 64.1%; p < 0.001). Patients with RVD at baseline had a higher estimated risk of death as assessed by Society of Thoracic Surgeons score and logistic EuroSCORE compared with patients with normal RV function. Baseline characteristics are summarized in Table 1.
Parameters from echocardiographic assessment at baseline are displayed in Table 2. Patients with RVD at baseline were found to have a lower left ventricular ejection fraction (LVEF) (45.1 ± 16.5% vs. 56.6 ± 12.5%; p < 0.001) and a lower mean transvalvular gradient (39.8 ± 17.7 mm Hg vs. 45.5 ± 17.3 mm Hg; p = 0.001). Patients with RVD more commonly had concomitant moderate or severe mitral (38.1% vs. 20.4%; p < 0.001) and tricuspid regurgitation (28.8% vs. 14.7%; p < 0.001), and higher mean pulmonary artery pressures (51.6 ± 15.0 mm Hg vs. 46.1 ± 14.1 mm Hg; p = 0.001) compared with patients with normal RV function.
Procedural characteristics are summarized in Table 3 and were comparable between patients with versus without RVD at baseline. Duration of hospitalization was, however, longer in patients with RVD compared with patients with normal RV function at baseline (9.8 ± 5.1 days vs. 8.7 ± 4.1 days; p = 0.001).
Short- and long-term clinical outcomes according to presence or absence of RVD are summarized in Table 4 and illustrated in Figure 3. Clinical outcomes in competing risk with death according to RV function are provided in Supplemental Table 4. Single imputation of missing data was performed before any of the adjusted analyses using the median or mode (missing data: n = 4 chronic obstructive pulmonary disease [assumed no chronic obstructive pulmonary disease], n = 6 body mass index [assumed above 20 kg/m2], n = 3 post-AR [assumed mild], n = 2 NYHA functional class [assumed III or IV]). MACCE occurred in 29.6% of patients with versus 14.7% of patients without RVD at 1 year (HRadj; 2.08; 95% CI: 1.57 to 2.76; p < 0.001). The difference was driven by a higher rate of all-cause mortality in patients with RVD compared with patients with normal RV function (26.2% vs. 11.1%; HRadj: 2.48; 95% CI: 1.82 to 3.38; p < 0.001). The higher mortality in patients with RVD emerged within the first 30 days after TAVR (9.9% vs. 2.7%; HRadj: 4.09; 95% CI: 2.33 to 7.18; p < 0.001) and was driven by a higher rate of cardiovascular death (9.0% vs. 2.2%; HRadj: 4.62; 95% CI: 2.51 to 8.50; p < 0.001). There was an inverse correlation of FAC and cardiovascular mortality at 1 year (HRadj: 0.95; p < 0.001) (Supplemental Table 5). In addition, patients with RVD had a higher rate of kidney injury stage 3 as compared with patients with normal RV function (HRadj: 2.43; 95% CI: 1.28 to 4.61; p = 0.007).
In a multivariate analysis, RVD was the strongest independent predictor of 1-year cardiovascular mortality (HRadj: 2.51; 95% CI: 1.64 to 3.86), followed by moderate or greater post-procedural AR (HRadj: 2.02; 95% CI: 1.18 to 3.46) (Table 5).
The effect of RVD on cardiovascular mortality was maintained in multivariable models including tricuspid regurgitation, pulmonary artery systolic pressure, or mitral regurgitation (Supplemental Table 6).
Post-procedural recovery of RV function
Raw echocardiographic data for the assessment of RV function after TAVR was available in 1,043 (93.5%) patients. Echocardiographic follow-up was performed after a median of 2 days after the index procedure (interquartile range: 0 to 20 days). Among 305 patients with RVD at baseline and echocardiographic data for an evaluation of post-procedural RV function, recovery of RV function was observed in 175 (57.4%) patients. In contrast, 89 (12.1%) patients were found to develop new RVD after TAVR. Baseline, echocardiographic, and procedural characteristics across categories of RVD pre- and post-TAVR are provided in Supplemental Tables 7 to 9. There was a gradient of increasing risk of cardiovascular death across categories of patients with normal RV function, RVD at baseline that recovered after TAVR (HRadj: 2.16; 95% CI: 1.16 to 4.02; p = 0.01), patients with newly developed RVD after TAVR (HRadj: 3.93; 95% CI: 2.09 to 7.39; p < 0.001), and patients with RVD at baseline that was maintained after TAVR (HRadj: 8.74; 95% CI: 5.33 to 14.3; p < 0.001) (Supplemental Table 10, Supplemental Figure 1). Predictors of RV recovery or RV deterioration are shown in Supplemental Table 11. Atrial fibrillation and lower LVEF were identified as predictors of RVD and negatively predicted recovery from RVD after the intervention, respectively.
The key findings of the present analysis of the effect of RVD on clinical outcomes after TAVR can be summarized as follows: 1) 29% of patients undergoing TAVR had RVD at baseline according to the most recent ASE/EACVI echocardiographic guidelines; 2) patients with RVD at baseline had a 2.9-fold (95% CI: 2.0- to 4.3-fold) increased risk of cardiovascular death at 1 year after TAVR and RVD emerged as the strongest independent predictor of mortality at 1 year; 3) the higher cardiovascular mortality emerged within the first 30 days after TAVR; 4) RVD tended to normalize in 57% of patients after TAVR; 5) there was a gradient of risk of death across categories of patients that increased from patients with normal RV function, over those with RVD recovery and new RVD, to those with persistent RVD.
In patients undergoing TAVR, RVD was recorded at baseline in 29% of patients using the echocardiographic guidelines of the ASE/EACVI (8). The documented prevalence was comparable to previous reports, most of which used alternative criteria for RV assessment (1–3,6). RVD may be a direct consequence of advanced AS or result from unrelated comorbidities, and resonates with the high-risk profile in this patient population. Long-standing elevated left-sided filling pressures can translate into pulmonary vascular remodeling and pulmonary arterial hypertension with ensuing RV impairment (1). At the same time, the risk of RV deterioration after cardiac surgery in patients with RVD at baseline may have inclined the heart team toward a transcatheter approach. The potential selection bias of patients with RVD in the TAVR cohort may be further amplified by the allocation of patients with advanced pulmonary disease to TAVR to forego the potential risk of ventilation problems.
RVD has previously been associated with mortality from any cause in patients with heart failure or patients undergoing cardiac surgery (14,15). More recently, several studies have addressed the effect of RV geometry and function, pulmonary hypertension, and tricuspid regurgitation on clinical outcomes after TAVR (5,16–19). We found a more than 2-fold increased risk of all-cause death in patients with RVD at 1 year that was robust after multivariable adjustment for patient comorbidities. Our findings are consistent with 2 previous reports from smaller patient cohorts that characterized RVD according to FAC or TAPSE, respectively. RV FAC was associated with increased mortality during long-term follow-up in a study of 539 patients undergoing left-sided valvular intervention. There was no significant interaction between RVD and left ventricular dysfunction (20). In an observational study of 702 patients undergoing TAVR, mild to moderate RVD as defined by TAPSE 14 to 18 mm and severe RVD as defined by TAPSE <14 mm were associated with an increased risk of long-term mortality (TAPSE 14 to 18 mm, HR: 1.53; 95% CI: 1.07 to 2.18; TAPSE <14 mm, HR: 2.12; 95% CI: 131 to 3.34) (6). In contrast, 2 reports showed no association of RVD with mortality. In a subanalysis of the inoperable cohort of the PARTNER II trial with RV function assessed by FAC in an independent core laboratory, RVD was not associated with an increased risk of mortality in a multivariable analysis (5). Another analysis in 606 patients indicated a numerically increased risk of mortality at 1 year in patients with RVD as assessed by the recommendations of the ASE; however, this did not reach statistical significance (3), and the analysis was not adjusted for differences in baseline characteristics. The conflicting results of different studies with regard to the effect of RVD on clinical outcome are most likely attributable to inconsistent definitions of RVD, modest study populations with limited duration of follow-up, and differences in the analytical approach.
The interplay between tricuspid regurgitation and RVD is equivocal. We found moderate or severe tricuspid regurgitation in 28.8% of patients with RVD. The effect of RVD on 1-year cardiovascular mortality was maintained after adjustment for tricuspid regurgitation in our analysis (HRadj: 2.61; 95% CI: 1.77 to 3.85). Moderate or severe tricuspid regurgitation irrespective of RVD has been associated with a 1.76× (95% CI: 1.14 to 2.70×) increased risk of 1-year mortality in a subanalysis of the PARTNER II trial; this contrasts with a study by Kammerlander et al. (20) indicating that RVD, but not tricuspid regurgitation, was significantly associated with survival after left-sided heart valve procedures (5).
RVD tended to normalize in more than one-half of all patients with RVD at baseline. In contrast to an analysis of 226 patients with RVD defined by TAPSE only, we found a lower risk of death in patients with recovery of RVD after TAVR compared with patients with maintained RVD (21). Unlike the previously mentioned study, we evaluated RV function using several parameters according to the latest guidelines and defined the recovery of RV function as a normalization of RV function, not an improvement of severity.
We observed a gradient of risk for death across categories of patients increasing from those with normal RV function, to those with RVD recovery, new RVD, and persistent RVD, respectively, suggesting that the evaluation of RV function pre- and post-TAVR may need to be considered in pre- and post-procedural risk assessment.
First, all subjects were recruited at a single center, and we were unable to evaluate all consecutive patients who underwent TAVR due to a lack of echocardiographic data. A total of 154 patients were excluded due to lack of echocardiographic data, and 50 patients were excluded due to inadequate quality of the echocardiographic clips. Second, follow-up echocardiographic data with full assessment of RVD was not present in all patients. Third, for secondary endpoints with low event rates, the precision of the effect size comparison of RVD versus normal RV function was low and interpretability was therefore limited. Finally, the possibility of residual confounding cannot be excluded. Notwithstanding, this is the first report to reveal the association between RVD and clinical outcomes after TAVR as assessed by the most recent echocardiographic guidelines.
Our findings have important implications for clinical practice. RVD defined according to the latest guidelines contributes to the identification of a substantial proportion of the routine TAVR population at increased risk of early and late mortality. The appropriate treatment strategy of these patients with particularly high risk of adverse clinical outcome need to be further evaluated in prospective clinical studies.
COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: Up to one-third of patients undergoing TAVR were found to have RVD at baseline, more than one-half of which recovered to normal RV function after TAVR. Compared with patients with normal RV function, we found a 2-fold increased risk of early and late mortality in patients with RVD before TAVR and a 6-fold increased risk of late mortality in patients with no recovery of RVD after TAVR.
TRANSLATIONAL OUTLOOK: RVD before and after TAVR may need to be considered in the risk stratification of patients undergoing TAVR.
Dr. Praz has served as a consultant for Edwards Lifesciences. Prof. Räber has received research grants to his institution from Biotronik, Sanofi, and Regeneron. Prof. Wenaweser has served as a proctor for Medtronic, Edwards, New Valve Technology, and Boston Scientific. Prof. Windecker has received research grants to his institution from Abbott, Amgen, Boston, Biotronik, and St. Jude Medical. Prof. Pilgrim has received research grants to his institution from Edwards Lifesciences, Symetis, and Biotronik; has received speaker fees from Boston Scientific; and has received reimbursement for travel expenses from St. Jude Medical. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- aortic regurgitation
- aortic stenosis
- American Society of Echocardiography
- fractional area change
- major adverse cardiac and cerebrovascular events
- right ventricle/ventricular
- right ventricular dysfunction
- tricuspid annular plane systolic excursion
- transcatheter aortic valve replacement
- Received October 30, 2017.
- Revision received December 18, 2017.
- Accepted December 21, 2017.
- 2019 American College of Cardiology Foundation
- Cavalcante J.L.,
- Rijal S.,
- Althouse A.D.,
- et al.
- Koifman E.,
- Didier R.,
- Patel N.,
- et al.
- Lindman B.R.,
- Maniar H.S.,
- Jaber W.A.,
- et al.
- Griese D.P.,
- Kerber S.,
- Barth S.,
- Diegeler A.,
- Babin-Ebell J.,
- Reents W.
- Harjola V.P.,
- Mebazaa A.,
- Celutkiene J.,
- et al.
- Wenaweser P.,
- Pilgrim T.,
- Kadner A.,
- et al.
- Kappetein A.P.,
- Head S.J.,
- Genereux P.,
- et al.
- Duncan A.,
- Ludman P.,
- Banya W.,
- et al.
- Ludman P.F.,
- Moat N.,
- de Belder M.A.,
- et al.
- Pilgrim T.,
- Englberger L.,
- Rothenbuhler M.,
- et al.
- Di Salvo T.G.,
- Mathier M.,
- Semigran M.J.,
- Dec G.W.
- D'Ascenzo F.,
- Conrotto F.,
- Salizzoni S.,
- et al.
- Ito S.,
- Pislaru S.V.,
- Soo W.M.,
- et al.
- Kempny A.,
- Diller G.P.,
- Kaleschke G.,
- et al.
- Kammerlander A.A.,
- Marzluf B.A.,
- Graf A.,
- et al.
- Testa L.,
- Latib A.,
- De Marco F.,
- et al.