Prognostic Impact of Different Subtypes of Severe Aortic Stenosis after Transcatheter Aortic Valve Implantation

Purpose Outcomes for different subtypes of aortic stenosis (AS) after transcatheter aortic valve implantation (TAVI) are still subjects of debate. The aim of the study was to evaluate the prognostic impact of baseline ow rate (F) and mean gradient (G) on survival after TAVI. Methods

Despite the favorable outcome and hemodynamic results after TAVI, a considerable proportion of patients die within the rst 2 years [11.]. Therefore, it is of particular importance to identify patients who may bene t from this treatment indeed. Surgical scores have not been validated to predict mortality after TAVI [12.]. Further risk scores have been developed, speci cally for TAVI patients [13.-14.], however, these systems are not widely accepted and used. Several studies revealed that reduced SVI may be an independent predictor of mortality after TAVI [15.,16.]. In addition, there have been similar conclusions regarding the impact of low ejection fraction (EF) and low gradient (LG) on outcomes after TAVI [17-19.]. Further extension of indication for TAVI can be expected in the near future in a signi cant part of low-risk AS patients, considering the positive results of the latest large low-risk TAVI trials [20.21.]. In this group long-term success is determined mostly by initial hemodynamic (i.e. echocardiographic) parameters and periprocedural complications rather than severe comorbidities.
The objectives of this study were to evaluate the prognostic impact of different SVI -gradient patterns on outcomes of TAVI, as well as the changes in left ventricular function.

Methods: Patients and Procedures
The present study included all (n 300) patients between 2008 and June 2017, with symptomatic (NYHA class II.-IV.) severe AS (mean Gr ≥ 40 mm Hg, or an aortic valve area of < 1.0cm, or < 0.6 cm 2 /m 2 ) undergoing TAVI in our Institute. All patients were evaluated for TAVI by the 'Heart Team'. In all patients, clinical assessment, transthoracic echocardiography, coronary angiography and multislice computed tomography were performed before the procedure.
The present study was accepted by the local ethics committee and all patients provided written informed consent.
Clinical and echocardiographic examinations of patients were performed within 7 days before TAVI and 12 months after implantation. All patients were followed for at least 3 years. Mortality tracking was achieved in 100% of patients. Mortality data were obtained in all cases from the Registry of the National Health Care System.
The primary endpoints were 30-day, 1-year and 3-year all-cause mortality after TAVI. In addition, secondary endpoint was the change of LVEF in the survivors.
Patients were subdivided into groups depending on their SVI -Gr pattern. High gradient was considered if the mean Gr was >/= 40 mmHg, and low Gr if it was < 40 mmHg. Low ow was de ned with SVI </= 35 ml/m 2 , and normal ow with SVI > 35ml/m 2 . Patients were divided into four groups: HG: high gradient; regardless of EF or SVI LF-LG: low ow -low gradient; and LVEF < 50% PLF-LG: low ow -low gradient and LVEF >/= 50% NF-LG: normal ow-low gradient.
Transcatheter aortic valve implantation TAVI was performed in 300 patients (native AS, n 294; degenerated bioprosthesis, n 6). Medtronic CoreValve prosthesis was inserted in 285 cases, Edwards-Sapient valve in 2 cases, Medtronic Engager in 8 cases and Lotus prosthesis in 7 cases. Access site for TAVI was transfemoral in 287 cases, transapical in 8 cases and direct transaortic in 5 cases. For each study, multiple transducer positions were used routinely for measurement of aortic gradient. Heart valve insu ciencies were evaluated according to the recommendations of the European Association of Echocardiography. Left ventricular ejection fraction was measured by biplane Simpson method from four-and two-chamber views.
All TAVI implantations were monitored with transesophageal echocardiography. Besides valve positioning, hemodynamics and possible complications, we carefully evaluated the cause and the degree of paravalvular aortic regurgitation (AR) according to the WARC-2 criteria [23.]. Measuring the circumferential extent of paravalvular jet(s) mild AR was de ned < 10%, moderate AR 10-29% and severe AR >/= 30%. Different grades of paravalvular AR are illustrated on Fig. 1..

Statistical analysis
Numerical values are expressed as mean + standard deviation (SD) or median. Continuous variables were compared between groups using an unpaired t-test (for normally distributed variables). Continuous variables were compared using the paired Student's t-test before and after TAVI. Time-to-event data analysis was done using the Cox proportional hazards model. Survival curves were constructed for timeto-event variables using Kaplan-Meier estimates and compared by the log-rank test. The association between mortality and baseline clinical, echocardiographic and periprocedural variables was assessed with the use of Cox proportional hazard analysis. Variables with a p value < 0.10 in univariable analysis were entered into the multivariable model. All reported probability values are two tailed, and P < 0.05 was considered statistically signi cant. Analyses were performed with the SPSS statistical software package (version 24.0).
Time to follow-up was median 34 months (0-115 months) The HG AS subgroup included 237 cases (79%), the LF-LG AS 41 cases (13,67%), while only 9 patients (3%) ful lled PLF-LG and 13 (4,33%) NF-LG criteria. Subgroups are illustrated in Table 1.  HG AS patients were signi cantly older, had smaller AVA,. From the 237 HG cases 127 (53,6%) had low SVI. In the LF-LG group the prevalence of NYHA class IV., myocardial infarction, peripheral artery disease was higher, and these patients had larger end-diastolic and end-systolic volumes and lower EF compared to HG-AS group. In the PLF-LG group the ESV was smaller and the EF was higher than in the HG patients.
In the NF-LG group the EDV and ESV was signi cantly higher than in the HG group. Baseline clinical and hemodynamic characteristics of patients and the differences between groups are shown on Table 2.   Fig. 3. Differences in survival between HG and LF-LG groups is demonstrated on Fig. 4.

Predictors of mortality:
Because of the course of mortality rates, we considered of importance the assessment of 1-year hazard ratios of different subgroups, as well as several degrees of EF, ow, Gr, AVA and comorbidities. Different degrees of left ventricular dysfunction and low gradient had signi cantly higher risk of mortality, while lower ow or area were not independently predictive. When considering different subtypes of AS, LF-LG AS had a more than twofold risk of mortality (HR 2,44 [1, 3-4, 42], p < 0,003) compared to HG. Out of the comorbidities, only patients with severe renal insu ciency (GFR < 30ml/min), or with moderate and severe paravalvular aortic regurgitation (AR) had higher risk of mortality, besides of advanced (80+) age. The hazard ratios are demonstrated on Table 4. With logistic regression analysis -independent predictor of 1-year all-cause mortality were moderate or severe paravalvular aortic regurgitation (AR) and severe renal failure (GFR < 30ml/min), while area, ow, EF, gradient, or other concomitant diseases were not statistically signi cant -as shown on Table 5. With the analysis of the interdependence of different echocardiographic parameters there was a moderate and positive correlation between ejection fraction and ow (R = 0,541, p < 0,0001), EF and gradient (R = 0,485, p < 0,0001), ow and area (R = 0,495, p < 0,0001), and an inverse but moderate correlation between area and gradient (R = 0,454, p < 0,0001), but the correlation of ow and gradient was weak (R = 0,206, p < 0,0001). These correlations are demonstrated on Fig. 5.
On the other hand, when classifying our patients solely by their initial EF, a clinically relevant and statistically signi cant improvement of EF could have been demonstrated in every ten-percent cluster under EF 60%, as demonstrated on Fig. 7. .

Discussion:
Patients with LF-LG AS represent a challenging subpopulation with a poor prognosis when undergoing conservative or surgical treatment. TAVI is an alternative option, however, due to controversial data, the outcomes of TAVI and predictors of late mortality have not been clari ed yet [24.]. Different risk algorithms were suggested for risk strati cation based on clinical and laboratory ndings and LVEF, but none of them used Gr or SVI for risk strati cation [13.-14.]. Since it is still uncertain whether these patients bene t from less invasive treatment, we aimed to evaluate the prognostic impact of different SVI-gradient AS patterns after TAVI. The main nding is that 30-day, 1-year and 3-year all-cause mortality was the highest in the LF-LG group. Our 1-year mortality results are comparable to larger registries [25.,26.]. The 30-day and 1-year mortality rate of the LF-LG subgroup was about 2-3 times higher than in the HG AS patients.
Reduced LVEF was found to be a predictive factor of 30-day and 1-year mortality after TAVI in several studies, [26.27.]. In a recent meta-analysis of 6898 TAVI patients low SVI, low gradient, and low EF were each associated with higher mortality after TAVI [28.].
In our study the PLF-LG group represented only 3,0% of the whole patient population. The 1-year mortality was worse than in patients with HG AS, but better than in the LF-LG group. We could not demonstrate the negative effect of low SVI, which is not surprising considering the high percentage (53,6%) of low ow condition in the HG group, in which the prognosis was better. On the other hand, we found moderately strong correlation between EF and ow, EF and gradient, ow and area and between area and gradient, so these parameters must not be emphasized solely. Beyond that, HG or LF-LG criteria were useful in the prediction of outcome. Our results correspond with one previous study in which AS subtype independently in uenced long-term mortality, whereas EF, ow or low gradient alone did not [29.].
In our study only moderate or severe paravalvular AR, and severe renal insu ciency were independent predictors of 1-year mortality. This nding is consistent with the results of larger registries [20.,30.]. Limitations: The main limitation of our study is the relatively smaller sample size, and its single-center character. We could not estimate the risk of the PLF-LG AS, since this subgroup was small, and only 3 of them reached primary endpoint at 1-year. Despite prospective data collection, the analysis was retrospective and therefore subject to the limitations of an observational study. Besides limitations, the strenghts of our study are the complete analysis of consecutive patients, the 100% mortality tracking, and the long followup period up to 115 months.

Conclusions:
Our study demonstrates the impact of different subtypes of AS on outcome after TAVI. All-cause mortality was lowest in patients with HG AS, it was intermediate in patients with PLF-LG AS and highest in patients with LF-LG AS. The classi cation of patients into ow-gradient pattern might be useful for risk strati cation before TAVI. Considering the high 1-year all-cause mortality in patients with LF-LG AS, a careful selection is warranted. In patients with LF-LG AS closer follow-up and active medical treatment of left ventricular dysfunction is needed after TAVI, in order to improve survival. Improvement of initial left ventricular dysfunction can be expected after TAVI regardless of its severity.

Declarations
Funding: None Con icts of interest: The authors declare that they have no con ict of interest.
Ethics approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the Institutional Ethics Committee and with the 1964 Helsinki Declaration and its later amendments. Ethical approval was waived by the local Ethics Committee in view of the retrospective nature of the study and all the procedures being performed were part of the routine care.
Consent to participate: All patients provided written informed consent Consent for publication Not applicable Availability of data and material Not applicable