Article Text
PDF
Abstract
Objective Pulmonary hypertension (PH) is associated with increased mortality after surgical or transcatheter aortic valve replacement (TAVR) for aortic stenosis (AS), and when the pulmonary artery pressure is particularly elevated, there may be questions about the clinical benefit of TAVR. We aimed to identify clinical and haemodynamic factors associated with increased mortality after TAVR among those with moderate/severe PH.
Methods Among patients with symptomatic AS at high or prohibitive surgical risk receiving TAVR in the Placement of Aortic Transcatheter Valves (PARTNER) I randomised trial or registry, 2180 patients with an invasive measurement of mean pulmonary artery pressure (mPAP) recorded were included, and moderate/severe PH was defined as an mPAP ≥35 mm Hg.
Results Increasing severity of PH was associated with progressively worse 1-year all-cause mortality: none (n=785, 18.6%), mild (n=838, 22.7%) and moderate/severe (n=557, 25.0%) (p=0.01). The increased hazard of mortality associated with moderate/severe PH was observed in females, but not males (interaction p=0.03). In adjusted analyses, females with moderate/severe PH had an increased hazard of death at 1 year compared with females without PH (adjusted HR 2.14, 95% CI 1.44 to 3.18), whereas those with mild PH did not. Among males, there was no increased hazard of death associated with any severity of PH. In a multivariable Cox model of patients with moderate/severe PH, oxygen-dependent lung disease, inability to perform a 6 min walk, impaired renal function and lower aortic valve mean gradient were independently associated with increased 1-year mortality (p<0.05 for all), whereas several haemodynamic indices were not. A risk score, including these factors, was able to identify patients with a 15% vs 59% 1-year mortality.
Conclusions The relationship between moderate/severe PH and increased mortality after TAVR is altered by sex, and clinical factors appear to be more influential in stratifying risk than haemodynamic indices. These findings may have implications for the evaluation of and treatment decisions for patients referred for TAVR with significant PH.
Trial registration NCT00530894.
Statistics from Altmetric.com
Introduction
Pulmonary hypertension (PH) is an emerging risk factor for increased mortality after surgical and transcatheter aortic valve replacement (TAVR).1–4 Patients with very elevated pulmonary pressures may be turned down for valve replacement due to concerns about high perioperative morbidity and mortality or questions about whether valve replacement will yield clinical benefit.5 There is uncertainty regarding how to further risk stratify this subgroup of patients, and what factors are associated with an acceptable versus poor clinical outcome. In this regard, pulmonary vascular resistance (PVR) or its reversibility in response to vasodilator challenge is, sometimes, considered to suggest whether the PH is reversible and to indicate the likelihood of clinical improvement.6 ,7 Whether this approach has validity is unknown.8 Accordingly, we evaluated the effect of clinical and haemodynamic factors on mortality among patients with moderate or severe PH undergoing TAVR in the Placement of Aortic Transcatheter Valves (PARTNER) I randomised trial and continued access registry. We hypothesised that a combination of clinical and haemodynamic factors would refine risk stratification of patients with significant PH, which could have important implications for treatment decisions of this higher-risk population.
Methods
Study population
The design, inclusion and exclusion criteria and primary results of the high-risk (cohort A) and prohibitive-risk (cohort B) cohorts of the PARTNER I randomised clinical trial have been reported.9 ,10 The inclusion and exclusion criteria for high-risk and prohibitive-risk patients enrolled in the continued access registry were the same as those enrolled in the randomised trial.9 ,10 Patients were symptomatic (New York Heart Association (NYHA) functional class ≥2), and had severe aortic stenosis (AS) with an aortic valve area (AVA) <0.8 cm2 (or indexed AVA <0.5 cm2/m2) and either resting or inducible mean gradient >40 mm Hg or peak jet velocity >4 m/s. High surgical risk was defined by a predicted risk of death of 15% or higher by 30 days after conventional surgery.10 Patients at prohibitive risk were not considered to be suitable candidates for surgery due to a predicted probability of death or a serious irreversible condition of 50% or higher by 30 days after conventional surgery.9 Based on an assessment of vascular anatomy, patients were deemed suitable for either a transfemoral or transapical approach and, if enrolled in the trial, randomised to transcatheter therapy with the Edwards SAPIEN heart valve system (Edwards Lifesciences, Irvine, California, USA) or surgical aortic valve replacement (cohort A) or transcatheter therapy or medical therapy (cohort B). For this analysis, we included only patients who received treatment with TAVR (the ‘as treated’ population) who also had an invasive measurement of pulmonary artery pressure recorded at the time of the TAVR procedure prior to balloon aortic valvuloplasty. Clinical characteristics were reported by the enrolling sites. The study protocol was approved by the institutional review board at each enrolling site, and all patients provided written informed consent.
Haemodynamics and echocardiography
A pulmonary artery catheter was routinely used during the TAVR cases to obtain invasive haemodynamic measurements before balloon aortic valvuloplasty and after transcatheter valve placement. The measurements obtained prior to balloon aortic valvuloplasty were used for this analysis. PH was defined as: any PH (mean pulmonary artery pressure (mPAP) ≥25 mm Hg), mild PH (mPAP 25 to <35 mm Hg), moderate PH (mPAP 35 to <45 mm Hg) and severe PH (mPAP ≥45 mm Hg). PVR=(mPAP−mean pulmonary capillary wedge pressure)/cardiac output. Pulmonary artery compliance=stroke volume index (determined by echocardiography)/pulmonary artery pulse pressure. An independent core laboratory analysed all echocardiograms as previously described; the baseline echocardiogram obtained within 45 days prior to TAVR was used in this analysis.11 ,12 Stroke volume was calculated as the LV outflow tract area multiplied by the pulsed wave Doppler LV outflow tract velocity-time integral and indexed to body surface area (BSA).
Clinical endpoints
Clinical events, including all-cause death, cardiac death, repeat hospitalisations, stroke, renal failure, major bleeding and vascular complications were adjudicated by a clinical events committee and defined previously.9 ,10 Disease-specific health status was assessed with the Kansas City Cardiomyopathy Questionnaire; the overall summary score was the primary health status outcome for the PARTNER trial.13–15
Statistical analysis
Continuous variables were summarised as mean±SD or medians and quartiles, as appropriate, and compared using analysis of variance for means or Kruskal–Wallis test for medians. Categorical variables were described as n (%) and compared using the χ2 test or Fisher’s exact test, as appropriate. Survival curves for time-to-event variables, based on all available follow-up data, were performed with the use of Kaplan–Meier estimates, and were compared between groups with the use of the log-rank test. Cox proportional hazards models were used to calculate HRs and to test for interactions. The proportional hazards assumption was checked and verified for each of the Cox models by creating interactions of the predictors and survival time, and none were found to be significant. Variables included in the multivariable Cox models were forced (age, sex) or had a univariable association (p<0.10) with 1-year all-cause death. To improve risk stratification among patients with moderate or severe PH, we evaluated the effect of having a progressive number of the factors that were independently associated with mortality in this population. Based on the ß-estimates for each of the independent predictors of mortality in the multivariable model, the variables were weighted, and a risk score was developed with scale 0–8. All statistical analyses were performed with SAS software V.9.2. A p value of <0.05 indicates evidence of a statistically significant effect.
Results
Patient population
Among 2180 patients with symptomatic AS at high or prohibitive surgical risk receiving TAVR in the PARTNER I trial or continued access registry, 436 (20%) were enrolled in the trial (cohort A, n=286; cohort B, n=150), and 1744 (80%) were in the continued access registry (cohort A, n=1508; cohort B, n=236). A transfemoral approach was used in 59% of patients, and the remainder were treated via a transapical approach. Clinical, echocardiographic and haemodynamic characteristics are shown in table 1. Higher pulmonary artery pressure was associated with higher body mass index, lower EF and indexed AVA, and a greater prevalence of diabetes, NYHA functional class 4 symptoms, major arrhythmia, obstructive lung disease and moderate-to-severe mitral and aortic regurgitation (table 1). In contrast, there was no association between pulmonary artery pressure and sex or the prevalence of systemic hypertension, smoking or renal impairment (table 1).
Baseline characteristics according to PH severity
Clinical outcomes at 30 days and 1 year
Higher pulmonary artery pressure was associated with a trend towards increased cardiovascular death (p=0.051) and stroke (p=0.052) at 30 days, but not with other clinical outcomes (table 2). One-year rates of all-cause death, cardiovascular death, repeat hospitalisation and stroke were higher in those with a higher pulmonary artery pressure (p<0.05 for all) (table 2). Although there were no differences in baseline NYHA functional class among PH groups, NYHA functional class was modestly worse at 30 days and 1 year in those with higher pulmonary artery pressures (figure 1A). Initially, health status was worse among patients with higher pulmonary artery pressures, but by 1 year, these differences were no longer observed (figure 1B). At baseline, those with mild PH or moderate/severe PH walked 20 m less in 6 min than those without PH (p≤0.005 for both), but at 30 days, there was no difference in 6 min walk distance between PH groups (figure 1C). At 1 year, there was a difference in 6 min walk distance between PH groups (p=0.008) owing to a shorter 6 min walk distance among those with mild PH compared with those without PH (p=0.002) (figure 1C).
Clinical outcomes based on pulmonary hypertension (PH)
Functional classification, health status and 6 min walk distance over time according to severity of pulmonary hypertension (PH). New York Heart Association (NYHA) functional class (A), Kansas City Cardiomyopathy Questionnaire (KCCQ) overall summary score (B) and 6 min walk distance (C) at different time points according to the severity of PH. The box and whisker plots show the median values, 25th and 75th percentiles (box), and the 2.5 and 97.5 percentiles (whiskers).
PH, mortality and sex
We evaluated whether the increased hazard of death associated with moderate/severe PH (compared with no PH) was consistent across subgroups. In females, moderate/severe PH was associated with increased mortality, whereas in males, it was not (interaction p=0.03) (figures 2 and 3). This interaction persisted after adjustment for factors associated with mortality (p=0.02) (table 3). In adjusted analyses, females with moderate/severe PH had an increased hazard of death at 1 year compared with females without PH (adjusted HR 2.14, 95% CI 1.44 to 3.18), whereas those with mild PH did not (table 3). Among males, there was no increased hazard of death associated with any severity of PH. Among patients with no or mild PH, males had an almost twofold higher adjusted hazard of death at 1 year, whereas there was no difference in mortality based on sex among those with moderate/severe PH (table 3).
Multivariable analysis evaluating the relationship between sex, pulmonary hypertension (PH) and 1-year all-cause mortality
Effect of clinical and echocardiographic factors on the association between moderate or severe PH and 1-year death from any cause. Cox proportional hazards models were used to evaluate the HRs for 1-year death from any cause for patients with moderate/severe versus no PH in the clinical and echocardiographic subgroups shown. COPD, chronic obstructive pulmonary disease; GFR, glomerular filtration rate; PH, pulmonary hypertension.
Sex, pulmonary hypertension (PH) and mortality. Time-to-event curves for all-cause death for risk scores based on sex and severity of PH. The event rates were calculated with the use of Kaplan–Meier methods and compared with the use of the log-rank test.
Mortality among patients with moderate/severe PH
We evaluated which clinical and haemodynamic factors were associated with mortality within the group of patients with moderate/severe PH (n=557). Among clinical factors with a univariable association with mortality, oxygen-dependent lung disease (HR 2.57, 95% CI 1.63 to 4.06, p<0.001), lower glomerular filtration rate (HR 1.07 per 5 mL/min/1.73 m2 decrease, 95% CI 1.01 to 1.14, p=0.03), lower aortic mean gradient (HR 1.07 per 5 mm Hg decrease, 95% CI 1.00 to 1.15, p=0.039) and inability to perform a 6 min walk (HR 1.66, 95% CI 1.15 to 2.39, p=0.007) were independently associated with increased 1-year mortality (table 4). After adjustment for clinical factors, none of the haemodynamic indices were associated with mortality, although there was a trend of an association between lower pulmonary artery compliance and increased mortality (p=0.07) (table 4). Although not a baseline variable that could be used in preoperative risk stratification, moderate/severe total aortic regurgitation (AR) on the discharge echocardiogram was associated with increased 1-year mortality among patients with moderate/severe PH (adjusted HR 1.95, 95% CI 1.51 to 2.54, p<0.001).
Multivariable analysis evaluating variables associated with 1-year mortality among those with moderate or severe pulmonary hypertension (PH)
Based on this multivariable analysis of patients with moderate or severe PH, we developed a simple risk score (range 0–8), including oxygen-dependent lung disease (3 points), glomerular filtration rate <40 mL/min/1.73 m2 (2 points), inability to perform a 6 min walk (2 points) and baseline aortic mean gradient <40 mm Hg (1 point) (see online supplementary table S1). The risk score had an area under the curve (AUC) of 0.65 (95% CI 0.60 to 0.71) for 1-year death following TAVR. Based on an receiver operating characteristic (ROC)-derived cut-point of ≥3 vs <3, the risk score had a sensitivity of 62.8%, specificity of 59.4%, positive predictive value (PPV) of 33.2% and negative predictive value (NPV) of 83.2% for 1-year death following TAVR. Among patients with moderate or severe PH and a low score (0–1), only 14.7% died by 1 year. In contrast, among those with a high score (6–8), 59.4% died by 1 year after TAVR (figure 4).
Risk score and mortality among patients with moderate or severe pulmonary hypertension. Time-to-event curves for all-cause death for risk scores 0–1, 2–3, 4–5 and ≥6. The event rates were calculated with the use of Kaplan–Meier methods and compared with the use of the log-rank test. The risk score was comprised of four variables with scale 0–8: oxygen-dependent lung disease (3 points), glomerular filtration rate <40 (2 points), 6 min walk (unable to perform) (2 points) and baseline transvalvular mean gradient <40 mm Hg (1 point).
Discussion
In patients with symptomatic AS at high or prohibitive surgical risk treated with TAVR, we found that increased pulmonary pressures are associated with increased mortality, more frequent repeat hospitalisations and more strokes during the first year after the procedure. The increased hazard of mortality associated with moderate/severe PH was observed in females, but not males. Haemodynamic factors were not helpful in further risk stratifying patients with moderate/severe PH. In contrast, clinical factors, including oxygen-dependent lung disease, inability to perform a 6 min walk, worse renal function and lower aortic mean gradient were independently associated with increased mortality after TAVR in patients with moderate/severe PH. A weighted combination of these factors was able to identify patients with a 15% vs 59% 1-year mortality. Further studies are needed to confirm and extend these findings to refine risk stratification of high-risk patients with significant PH evaluated for TAVR as this may have important implications for treatment decisions.
PH is more common than often recognised in patients with AS, present in up to 65% of patients with severe symptomatic AS.1 ,16 ,17 Exercise-induced PH is present in over half of patients with severe asymptomatic AS and associated with increased cardiac events.18 A significant proportion of patients with PH have an elevated PVR, which further drives up the pulmonary pressures beyond those with only passive pulmonary venous hypertension.7 ,17 Several studies have demonstrated the increased hazard of mortality associated with PH in patients with AS treated with surgical valve replacement or TAVR.1–4 Some patients with severe PH are turned down for surgery, which has a dismal prognosis.19 Although pulmonary pressures tend to decrease after aortic valve replacement, some patients with severe preoperative PH have a persistent severe elevation in pulmonary pressures postoperatively, which is associated with a higher mortality than if the pulmonary pressures decrease.20 Given the increased hazard of death associated with significant PH, there is, sometimes, uncertainty in high-risk patients referred for TAVR regarding anticipated clinical benefit from valve replacement.5 Our findings confirm and extend prior studies in this area by identifying an interaction between PH and sex with respect to mortality after TAVR and identifying clinical factors as more important in risk stratifying patients with significant PH than haemodynamic factors.
Sex and PH
A novel finding from our study was the interaction between PH severity and sex with respect to mortality. The increased hazard of death associated with moderate/severe PH was only observed in females. Importantly, however, females with moderate/severe PH did not have worse outcomes than males. Rather, females with no or mild PH had much better survival than males, and only when the PH was moderate or severe were outcomes similar between the sexes. Why was an increase in pulmonary artery pressures associated with increased mortality among females, but not males? In light of the numerous studies demonstrating an association between increasing pulmonary pressures and increased mortality after aortic valve replacement, the lack of this association in males is the more unexpected finding. Given the age of this patient cohort, hormonal differences are an unlikely explanation. Unlike pulmonary arterial hypertension, which has a female predominance, we did not observe any association between sex and PH severity (table 1). Perhaps other clinical factors are more powerful drivers of mortality in males such that the adverse effects of PH are diluted by other contributing factors. Further studies are needed to confirm this interaction between PH and sex with respect to mortality and elucidate mechanisms for it.
Haemodynamic indices not helpful in risk stratification
When evaluating a patient with valvular heart disease and significant PH, there may be uncertainty regarding whether the proposed valve procedure will yield the desired symptomatic and functional benefit, which is often related to whether the PH is thought to be ‘reversible’.5 The severity of elevation in PVR and perhaps the decline in pulmonary pressures or PVR after the administration of a vasodilator may be interpreted as suggesting the reversibility of the PH and likelihood of clinical benefit from a mitral or aortic valve procedure.6 ,7 However, the specific drug used in this ‘vasodilator challenge’ may influence the haemodynamic response, and data are lacking regarding the conclusions to be drawn about the anticipated clinical benefit of valve surgery based on preoperative PVR elevation or vasodilator responsiveness.8 ,21 ,22 Our study did not evaluate acute pre-TAVR reversibility of PH or elevated PVR or the change in pulmonary pressures over time after TAVR. However, our observations deemphasise the importance of haemodynamic indices in further risk stratifying patients with significant PH being evaluated for TAVR. Interestingly, of all haemodynamic indices evaluated, pulmonary artery compliance (as an index of vascular stiffness) appeared to have greater prognostic utility than other measurements, which is consistent with studies in pulmonary arterial hypertension cohorts.23
Clinical implications
All patients with significant PH should not be considered alike in terms of the risk of adverse post-TAVR clinical outcomes. Rather than focusing on ways to risk stratify these patients based on haemodynamics, clinical factors appear to be more helpful in predicting mortality. Accordingly, instead of trying to risk stratify based on more detailed assessment of the pulmonary vasculature, it is more helpful to put the PH in the context of other organ dysfunction, including the lungs, kidneys and cardiac performance. The constellation of factors in the clinical risk score identifies those at particularly high risk for a poor outcome. When none/few of these factors were present, 1-year mortality was low (15%), whereas 1-year mortality was quite high (59%) when multiple factors were present, which may indicate the futility of TAVR in these patients. Among those with significant PH preoperatively, the clinical impact of persistent significant PH versus resolution/reduction of PH after TAVR requires further study as does the potential benefit of PH-directed therapy (eg, phosphodiesterase type 5 inhibitors) in these patients.7
Limitations
These findings apply to older patients with symptomatic AS at high surgical risk, and may not translate to younger, lower-risk patients. We did not have echocardiographic measurements of pulmonary pressures at baseline or during follow-up, so, we were unable to evaluate the effect of post-TAVR PH on clinical outcomes. The pulmonary pressures used in this study were obtained during the TAVR procedure (although prior to valvuloplasty or valve implantation) when patients were under general anaesthesia, likely lowering the pressures. However, invasive haemodynamics were more accurate than those obtained by an echocardiogram, and anaesthesia was likely to affect patients in the study similarly. We did not have the heart rate recorded at the time of the invasive haemodynamics, so we had to use the stroke volume obtained on the baseline echocardiogram to calculate pulmonary arterial compliance. This method has shortcomings in the presence of concomitant regurgitant lesions of the mitral and aortic valves. Measurements of RV function were not obtained in the PARTNER I cohort, which prevented us from looking at RV—pulmonary arterial coupling in this study. Finally, the risk score reported for mortality has not been validated, and only applies to patients with moderate or severe PH according to an invasive measurement of pulmonary artery pressure. Our purpose was not to develop a widely used risk score, but simply to demonstrate the broad range of outcomes in patients with moderate or severe PH based on the presence of these particular clinical factors.
Conclusion
PH is associated with lower survival after TAVR. Among those with moderate or severe PH, clinical factors are helpful in risk stratification, whereas haemodynamic factors appear to be less informative. TAVR may be futile for patients with significant PH with concomitant oxygen-dependent lung disease, impaired renal function, mobility impairment and low aortic valve gradients. Further studies are needed to elucidate the relationship between PH and sex with respect to mortality after TAVR, the effect of post-TAVR PH on clinical outcomes and whether PH-targeted therapies might improve clinical outcomes in these higher-risk patients.
Key messages
What is already known on this subject?
Pulmonary hypertension (PH) is associated with increased mortality after surgical or transcatheter aortic valve replacement (TAVR) for aortic stenosis.
What might this study add?
This study indicates that clinical factors are more helpful than haemodynamic indices in stratifying risk among patients with moderate or severe PH evaluated for TAVR. A weighted combination of clinical factors was able to identify patients with a 15% vs 59% 1-year mortality. Moreover, the association between moderate/severe PH and increased mortality after TAVR was only observed in females.
How might this impact on clinical practice?
TAVR may be futile for patients with significant PH with concomitant oxygen-dependent lung disease, impaired renal function, mobility impairment and low aortic valve gradients. Further studies are needed to elucidate the relationship between PH and sex with respect to mortality after TAVR, the effect of post-TAVR PH on clinical outcomes and whether PH-targeted therapies might improve clinical outcomes in these higher-risk patients.
Acknowledgments
The authors thank Maria Alu for her assistance in the preparation of the manuscript.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Data supplement 1 - Online supplement
Footnotes
Contributors DCM, JW, LGS, MBL and MJM contributed to the design of the PARTNER Trial. BRL, AZ, HSM and MJM contributed to the design of this analysis. BRL was responsible for drafting the manuscript. KX and GMA conducted the statistical analysis. All authors contributed to data acquisition and interpretation, provided substantive critical revision, have approved the final version and agree to take full responsibility for the accuracy and integrity of the work.
Funding The PARTNER trial was funded by Edwards Lifesciences, and the protocol was designed collaboratively by the Sponsor and the Steering Committee. The present analysis was carried out by academic investigators through the PARTNER Publications Office with no direct involvement of the sponsor in the analysis, drafting of the manuscript or the decision to publish. BRL is supported by the National Institutes of Health (grant number K23 HL116660).
Competing interests BRL is a site coinvestigator for the PARTNER Trial, and has received a research grant and serves on the scientific advisory board for Roche Diagnostics. AZ is a member of the PARTNER Trial Steering Committee, site PI for the PARTNER Trial and a consultant for Edwards Lifesciences. DCM has received consulting fees/honoraria from Abbott Vascular, St Jude Medical and Medtronic, and is an unpaid member of the PARTNER Trial Executive Committee. RMS is a member of the Steering Committees of the COAPT Trial (Abbott Vascular) and the PORTICO Trial (St Jude), national PI of the PERCEVAL Trial (Sorin Medical), has patent applications with Sorin and has received research grants from Sorin, Edwards, Abbott and St Jude. JW is a consultant for Edwards Lifesciences and an unpaid member of the PARTNER Trial Executive Committee. LGS holds equity in Cardiosolutions and ValvXchange, and has intellectual property rights/royalties from Posthorax, and is an unpaid member of the PARTNER Trial Executive Committee. SK has received consulting fees from Edwards Lifesciences, and is a member of the Scientific Advisory Board of Thubrikar Aortic Valve. VHT is a member of the PARTNER Trial Steering Committee and a consultant for Edwards Lifesciences, Sorin Medical, St Jude Medical and DirectFlow. SL has received consulting fees from Boston Scientific, Guerbet and St Jude Medical. MBL and MJM are unpaid members of the PARTNER Trial Executive Committee. The other authors report no potential conflicts of interest.
Ethics approval The institutional review board at each participating site approved the study.
Provenance and peer review Not commissioned; externally peer reviewed.