B‐type natriuretic peptide is associated with remodeling and exercise capacity after transcatheter aortic valve replacement for aortic stenosis

Background We aimed to assess longitudinal changes of B‐type natriuretic peptide (BNP) in aortic stenosis (AS) patients treated by transcatheter aortic valve replacement (TAVR). Methods From our TAVR database, we identified 193 consecutive patients with severe symptomatic AS who underwent TAVR and were prospectively followed using serial BNP levels and echocardiography. Patients were divided into subgroups according to type of left ventricular (LV) remodeling as having normal LV mass and relative wall thickness, or showing concentric remodeling (CR), concentric hypertrophy (CH), and eccentric hypertrophy (EH). Results At baseline, 30 patients (16%) had EH, 115 (60%) had CH, 37 (19%) had CR, and 11 (6%) had normal LV geometry. After TAVR, BNP decreased in the first 30 days, with further improvement during follow‐up. Patients with EH had higher BNP at baseline (P < 0.01) and a greater subsequent decrease (P < 0.001). During the median follow‐up of 1331 days (interquartile range: 632‐1678), 119 (62%) patients died. BNP showed a time‐dependent association with all‐cause mortality both in a univariable (hazards ratio [HR] 1.24, 95% confidence interval [CI]: 1.04‐1.47, P = 0.017), and in a multivariable model with Society of Thoracic Surgeons score and baseline BNP forced into the analysis (HR 1.32, 95% CI: 1.001‐1.73, P = 0.049). Elevated BNP was associated with a larger LV end‐diastolic volume index (P < 0.001) and shorter 6‐minute walk test distance (P = 0.013) throughout follow‐up. Conclusion In patients with AS, BNP was associated with LV remodeling phenotypes and functional status before and after TAVR. Elevated BNP levels were associated with poor prognosis.


| INTRODUCTION
B-type natriuretic peptide (BNP) levels are associated with disease severity, left ventricular (LV) remodeling, and prognosis in severe aortic stenosis (AS). 1,2 While studies have shown the prognostic utility of BNP at baseline and early after transcatheter aortic valve replacement (TAVR) in severe AS patients who underwent TAVR, 3-10 the factors that drive BNP change after AS correction by TAVR remain unclear.
In the present study, we aimed to assess the magnitude of longitudinal change of BNP in severe AS patients treated by TAVR, its impact on survival, and its association with functional status and underlying LV mechanics. We also assessed the determinants of BNP change and its association with specific LV remodeling patterns.

| Study sample
We reviewed our series of 237 consecutive high risk or surgically inoperable patients with severe AS who underwent TAVR through transfemoral access with the Edwards SAPIEN valve at Cleveland Clinic between May 2006 and December 11, 2012. 11 This included patients who enrolled in clinical trials (ie, REVIVAL, PARTNER trial, and PARTNER II) and patients who underwent commercial TAVR.
Plasma BNP measurements and echocardiographic assessments were done in a prospectively determined sequence at baseline, before discharge, 1 month, 6 months, 1 year, and annually thereafter. Baseline and subsequent echocardiographic studies were used to assess the association between BNP change and cardiac reverse remodeling after TAVR. We also evaluated the New York Heart Association (NYHA) functional class and 6-minute walk test (6MWT) results at the same point. The survival status of all patients after TAVR was collected from medical records or publicly available online sources (last queried April 2017). Patients were followed through chart review with either the date of last follow-up or the date of death recorded. We considered all-cause mortality to be the primary outcome. The secondary end point was the rate of death from cardiovascular causes. Cardiovascular deaths were defined as deaths resulting from a cardiac cause including sudden unexplained deaths in which a cardiac cause could not be excluded. 12 The study protocol was approved by the Cleveland Clinic Institutional Review Board, patient informed consent was waived and data were deidentified.

| Echocardiographic measurement
Baseline and subsequent echocardiographic measurements were systematically reviewed and measured by an experienced reader according to the current guidelines blinded to clinical, other variables, and outcome. 13  GLS was assessed by two-dimensional speckle-tracking echocardiography as previously reported. 11 The peak systolic strain values from apical four-chamber, two-chamber, and long-axis views were averaged to obtain GLS.

| Functional capacity
In the 156 patients who were enrolled in PARTNER trial and PARTNER II trial, 6MWT was conducted as part of prospectively determined protocol at baseline, 30 days, 6 months, and 1 year after TAVR. In nine patients, the 6MWT assessment was missed at the baseline and these patients were excluded from the analysis. Of note, in 49 patients, 6MWT was scheduled but not completed because of medical reasons (ie, postural hypertension, postural change in heart rate, low resting systolic pressure, patient being non-ambulatory because of medical condition, such as presence of angina or respiratory insufficiency at rest) and they were coded as zero meters. Data were obtained from research database.

| Statistical analysis
Continuous data are expressed as mean ± SD when normally distributed, or median (interquartile range). Categorical data are presented as an absolute number and percentages. BNP values were logarithmically transformed prior to analysis. We used the unpaired t test, Mann-Whitney test, One-way ANOVA, and Kruskal-Wallis test to compare the data between two or more groups as appropriate. To assess changes in BNP after TAVR over time, and possible impact of LV remodeling pattern, we applied longitudinal data analysis using a mixed effect model with unstructured covariance for random and fixed effects, 16,17 and with individual patients treated as random effects. Both BNP and time were logarithmically transformed. 11 Model selection was accomplished using log-likelihood ratio testing.
We also used a mixed effect model to assess the relationship between BNP levels and LVEDVi or 6MWT. To assess impact of time-varying effects on BNP survival after TAVR, we constructed a Cox proportional hazards model using BNP levels after TAVR as a segmented time-dependent covariate as described before. 18 Serial BNP levels were adjusted for difference in each time period (0, 1-10 days, 11-180 days, 181-450 days, >450 days) and added into the model as a time-dependent covariate. After constructing initial univariable model with BNP as a time-dependent covariate, we constructed the multivariable model by forcing baseline BNP and STS score. These variables were selected because of their known prognostic value in AS. 19 The assumption of proportional hazards were assessed based on the Schoenfeld residuals. A P-value of <0.05 was considered statistically significant. All statistical analyses were performed with JMP

| RESULTS
We identified 193 patients who fulfilled inclusion criteria (Table S1).
Six patients were excluded because of lack of baseline BNP assessment while 38 patients were excluded as they were followed by N-terminal pro BNP levels. "Commercial" TAVR was performed in 29 patients. A total of 30 patients (16%) were classified as having EH, 115 patients (60%) had CH, 37 patients (19%) had CR, and 11 patients (6%) were found to have normal LV geometry. Baseline EF and GLS were significantly reduced in patients with EH compared to patients with CH, CR, and normal patients (Table 1). In a next step, we divided patients into three groups according to their survival: patients who died during first 450 days, patients who died between 450 and 1200 days, and patients who survived for more than 1200 days. All three groups experienced a similar drop in BNP after TAVR (interaction between groups with time: P = 0.07). This indicates that decrease in BNP during follow-up was not related to the informative censoring (ie, caused by elimination through earlier  Figure 2).

| Factors associated with BNP levels after TAVR
BNP levels were associated with LVEDVi over the follow-up period (P < 0.001, Figure 3A). We also show a parallel decrease of BNP and LVEDVi during follow-up, suggesting that changes in LVEDVi and in BNP after TAVR coincide without significant time lag between the two. There was significant association between BNP improvement and GLS over time after TAVR (P < 0.001) (Supplemental Figure 3). Suggesting improvement in LV contractility is coincide with BNP change.
To assess the association between functional class and BNP levels, we collected NYHA class and 6MWT results. At baseline, 49 patients were unable to walk because of medical conditions and 6MWT distance was considered as 0 m. After TAVR, among those 49 patients, 19 patients were able to walk. The mean distance of the 6MWT was 222 ± 116 m at baseline and improved to 264 ± 121 m at 30 days after TAVR, followed by further subsequent improvement to 271 ± 124 m at 6 months after TAVR (P = 0.010) ( Table 2). In a mixed effect model, BNP levels showed significant negative correlation with 6MWT distance at baseline and over the follow-up period.
Significant parallel improvement in 6MWT distance and ln BNP level were observed (P < 0.001, Figure 3B). Similarly, NYHA class was significantly improved from 3.0 ± 0.4 to 2.1 ± 0.6 at 30 days after TAVR and also showed further improvement to 1.6 ± 0.7 at 1 year after TAVR (P < 0.001).

| DISCUSSION
In the present study, we explored the interconnection between processes of reverse remodeling, improvement in BNP levels, and increased exercise capacity that concomitantly occur after TAVR.
Overall, we observed that after TAVR patients with adverse LV remodeling show a continuing decrease in BNP during follow-up. This improvement in BNP coincided with recovery of cardiac structure and function. However, improvement in BNP was much less noticeable in patients without eccentric remodeling. The 6MWT distance, a marker of exercise capacity, also improved after TAVR, with ability to walk a longer distance again coinciding with lower BNP levels. The timedependent changes in post-procedural BNP levels were associated with all-cause mortality after TAVR.

| Association of BNP levels and LV remodeling pattern after TAVR
As it is known that BNP release is mediated by increased diastolic myocardial wall stress, 20 26 It also appears that the 6MWT is a useful surrogate of maximum exercise capacity in patients with heart failure and preserved ejection fraction. 27 Finally, recent data show that improvement in the 6MWT following TAVR is associated with better survival.. 28

| Clinical implication
In the present paper, we demonstrate that BNP levels after TAVR reflect the dynamic changes in cardiac structure and function. We first show that TAVR leads to early drop in BNP. We also show that there is a further continuing fall in BNP levels after TAVR, which is associated with improved LV structure and functional capacity. On the other side, we also show that a subsequent increase in BNP during followup is associated with increased chance of all-cause mortality. These findings suggest that BNP provides a mechanistic link between LV remodeling pattern and improved functional capacity in AS patients.
Hence, elevated BNP can serve as a surrogate of unfavorable clinical outcome after TAVR. BNP may be an inexpensive, convenient marker of post-TAVR cardiac remodeling, exercise capacity, and survival.

| Limitations
This was a retrospective observational study conducted at a large tertiary referral center and thus may suffer from selection bias. In the present study, only Edwards SAPIEN and SAPIEN XT valves were used, and the majority of patients underwent TAVR with the first generation SAPIEN valve. Moreover, although data were collected prospectively and analyzed by a mixed effect model, half of the patients died during the first 5 years of follow-up, which may result in survivorship bias.

| CONCLUSIONS
BNP levels were associated with a LV remodeling phenotype and functional status pre-and post-TAVR. Regression of BNP levels was prominent in patients with EH. Decrease of BNP levels was also associated with improved functional capacity. Elevated BNP levels after TAVR were independently associated with poor prognosis. In patients who underwent TAVR, serial follow-up of plasma BNP levels is useful in estimating prognosis.